Driver Behaviour anD Training

Pr

oo

fC

op

y

Driver Behaviour and Training

9781472414694_Dorn.indb 1

5/30/2013 10:19:49 AM

Human Factors in Road and Rail Transport Series Editors

y

Dr Lisa Dorn Director of the Driving Research Group, Department of Human Factors, Cranfield University

op

Dr Gerald Matthews Professor of Psychology at the University of Cincinnati

fC

Dr Ian Glendon Associate Professor of Psychology at Griffith University, Queensland, and President of the Division of Traffic and Transportation Psychology of the International Association of Applied Psychology

oo

Today’s society must confront major land transport problems. The human and financial costs of vehicle accidents are increasing, with road traffic accidents predicted to become the third largest cause of death and injury across the world by 2020. Several social trends pose threats to safety, including increasing car ownership and traffic congestion, the increased complexity of the human-vehicle interface, the ageing of populations in the developed world, and a possible influx of young vehicle operators in the developing world.

Pr

Ashgate’s ‘Human Factors in Road and Rail Transport’ series aims to make a timely contribution to these issues by focusing on the driver as a contributing causal agent in road and rail accidents. The series seeks to reflect the increasing demand for safe, efficient and economical land-based transport by reporting on the state-of-the-art science that may be applied to reduce vehicle collisions, improve the usability of vehicles and enhance the operator’s wellbeing and satisfaction. It will do so by disseminating new theoretical and empirical research from specialists in the behavioural and allied disciplines, including traffic psychology, human factors and ergonomics. The series captures topics such as driver behaviour, driver training, in-vehicle technology, driver health and driver assessment. Specially commissioned works from internationally recognised experts in the field will provide authoritative accounts of the leading approaches to this significant real-world problem.

9781472414694_Dorn.indb 2

5/30/2013 10:19:49 AM


op

y

Volume VI

Edited by

Pr

oo

fC

Lisa Dorn and Mark Sullman Cranfield University, UK

9781472414694_Dorn.indb 3

5/30/2013 10:19:49 AM

© Lisa Dorn, Mark Sullman and the contributors 2013 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior permission of the publisher. Lisa Dorn and Mark Sullman have asserted their right under the Copyright, Designs and Patents Act, 1988, to be identified as the editors of this work.

op

y

Published by Ashgate Publishing Limited Ashgate Publishing Company Wey Court East 110 Cherry Street Union Road Suite 3-1 Farnham Burlington, VT 05401-3818 Surrey, GU9 7PT USA England www.ashgate.com

fC

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library

Pr

oo

The Library of Congress Control Number: 2003058287

ISBN 9781472414694 (hbk) ISBN 9781472414700 (ebk – PDF) ISBN 9781472414717 (ebk – ePUB)

9781472414694_Dorn.indb 4

5/30/2013 10:19:49 AM

Contents List of Figures List of Tables Preface

ix xiii xvii

y

Part 1 Driver Education: The Role of Experience and Instruction Anticipation, Neural Function and Mastering Driving Timo Järvilehto, Veli-Matti Nurkkala, Kyösti Koskela and Jonna Kalermo

2

Does Driving Experience Delay Overload Threshold as a Function of Situation Complexity? Julie Paxion, Catherine Berthelon and Edith Galy

3

Risk Allostasis: A Simulator Study of Age Effects Britta Lang, Andrew M. Parkes and Michael Gormley

4

Development and Evaluation of a Competence-based Exam for Prospective Driving Instructors Erik Roelofs, Maria Bolsinova, Marieke van Onna and Jan Vissers

fC

op

1

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

3

13 25

43

Part 2 Driver Behaviour and Driver Training 5

Identifying the Characteristics of Risky Driving Behaviour 63 Christian Gold, Thomas Müller and Klaus Bengler

6

The Impact of Frustration on Visual Search and Hazard Sensitivity in Filmed Driving Situations Peter Chapman and Jodie Walton

75

Anger and Prospective Memory While Driving: Do Future Intentions Affect Current Anger? Amanda N. Stephens, Gillian Murphy and Steven L. Trawley

89

7

9781472414694_Dorn.indb 5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:49 AM


vi

8

Emotion Regulation of Car Drivers by the Physical and Psychological Parameters of Music Rainer Höger, Sabine Wollstädter, Sabine Eichhorst and Laura Becker

9

The Relationship between Seat Belt Use and Distracted Driving M. Eugènia Gras, Francesc Prat, Montserrat Planes, Sílvia Font-Mayolas and Mark J.M. Sullman

1

101 2

109

Self-evaluation Bias in Stopping Behaviour whilst Driving Ai Nakamura, Kan Shimazaki and Toshiro Ishida

117

11

Predicting the Future Driving Style of Novice Drivers: The Role of Self-evaluation and Instructors’ Ratings Following Driver Training Laura Šeibokaitė, Auksė Endriulaitienė, Rasa Markšaitytė, Kristina Žardeckaitė-Matulaitienė and Aistė Pranckevičienė

127

Improving Safety during Work-related Driving among Postal Van Drivers Simo Salminen

139

op

y

10

fC

12

13 14

oo

Part 3 Road Environment, In-Vehicle Technology and Driver Behaviour Evaluation of Visual Overtaking Distance Using a Driver’s Psycho-emotional Response Atis Zarins, Janis Smirnovs and Liga Plakane

15

Driver Fatigue Systems – How do they Change Drivers’ Behaviour? Katja Karrer-Gauss and Pawel Zawistowski

16

Ergonomics of Parking Brake Application: An Introduction Valerie G. Noble, Richard J. Frampton and John H. Richardson

17

The Compatibility of Energy Efficiency with Pleasure of Driving in a Fully Electric Vehicle Lena Rittger and Marcus Schmitz

9781472414694_Dorn.indb 6

151

Cognitive Distractions and their Relationship with the Driver 161 Oscar W. Williamson, Alan R. Woodside and Jonathan R. Seymour

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

179 187

207

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM

Contents

18

Learning from Accidents: Using Technical and Subjective Information to Identify Accident Mechanisms and to Develop Driver Assistance Systems Stefanie Weber, Antonio Ernstberger, Eckart Donner and Miklós Kiss

223

Part 4 Methodological Considerations in Measuring Driver Behaviour The Consistency of Crash Involvement Recall across Time James Freeman, Anders af Wåhlberg, Barry Watson, Peter Barraclough, Jeremy Davey and Mitchell McMaster

20

Controlling for Self-reported Exposure in Traffic Accident Prediction Studies Anders E. af Wåhlberg

22

op

21

The Wrong Tool for the Job? The Predictive Powers of the DBQ in a Sample of Queensland Motorists James Freeman, Peter Barraclough, Jeremy Davey, Anders af Wåhlberg and Barry Watson

fC

23

Predictive Validity and Cross-cultural Differences in the Self-reported Driving Behaviour of Professional Driver Students in Ecuador Daniela Serrano, María Sol Garcés and Luis Rodríguez Psychometric Properties of the Driving Cognitions Questionnaire in a Polish Sample Agata Blachnio, Aneta Przepiórka and Mark J.M. Sullman

Index

9781472414694_Dorn.indb 7

233

y

19

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

vii

247

261

277

293 303

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM

y op fC oo Pr 9781472414694_Dorn.indb 8

5/30/2013 10:19:50 AM

List of Figures

3.3 3.4 3.5 3.6

y

op

3.1 3.2

Change in the structure of action during training Driving simulator Predictors of subjective workload Effects of driving experience on subjective workload Predictors of the number of collisions Effects of subjective workload on the number of collisions Effects of subjective workload on the number of collisions depending on situation complexity The TRL DigiCar and instructor station Mean ratings and standard deviations for ratings of task difficulty, feeling of risk and likelihood of a crash on urban roads and on the dual carriageway Estimated marginal means plots for the significant interaction between speed and age on urban roads Estimated marginal means plots for the significant interaction between risk and age on urban roads Estimated marginal means plots for the main effect of age on urban roads Estimated marginal means plots for the main effect of age on the dual carriageway Model of competent task performance Cut-off scores and standard errors for 13 versions of the Theory of Driving Test Cut-off scores and standard errors for 15 versions of the Theory of Lesson Preparation Test Cut-off scores and standard errors for 15 versions of the Theory of Instruction and Coaching Test Results of the pilot study Mean risk scores of situations Average DPQ-scores Ratings and skin conductance measures during dangerous and safer sections of hazard videos divided by whether the participant had previously been frustrated by attempting to solve impossible anagrams or not (control condition) Physiological and eye movement measures during dangerous and safer sections of hazard videos divided by whether the participant had previously been frustrated by attempting to solve impossible anagrams or not (control condition)

fC

1.1 2.1 2.2 2.3 2.4 2.5 2.6

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

4.1 4.2

4.3

4.4

5.1 5.2 5.3 6.1

6.2

9781472414694_Dorn.indb 9

5 15 16 17 17 18 19 29 31 34 35 36 37 48 53 54 55 68 71 72

81

83

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM


x

8.4 10.1 10.2 10.3 10.4 10.5 10.6

oo

12.1 13.1 13.2 13.3 13.4 13.5 13.6 14.1

y

8.2 8.3

op

8.1

Self-reported anger across baseline, maintenance and implementation phases 95 Anger inducing scenes A: traffic jam B: slow moving car C: tailgater 102 Experimental order of events 103 Mean rated intensity of anger, type of music and music presentation 104 Amount of anger reduction in relation to the experimental conditions 106 Shape of the crossing and angle of the video camera 119 A single frame from a video 120 Evaluations from the other’s point of view and from one’s own point of view 121 Scatter plot of the evaluations from the other’s point of view and from one’s own point of view 121 Model of behaviour and self-evaluation 123 Distribution of the full stop rate before and after drivers were informed 124 Flow chart of the discussion process 142 Components of visual overtaking distance 153 Typical record of overtaking episode 154 Distribution of time and distance used for overtaking manoeuvre 155 Distribution of overtaking distance (S1+S3) 155 Distribution of overtaking speed (V1) 156 Samples of GSR artefacts 157 Mobile subscriptions, world and development level (ITU, 2011: 2) 164 Mobile phone ownership in Northern Ireland (CSU, 2011: 4) 165 Lane position for both visual and auditory cognitive tasks (Jamson and Merat, 2005: 91) 169 Task-Capability Interface model (Fuller, 2005: 465) 171 Levels of awareness and activity control loops (Bellet et al., 2009: 1208) 172 Tracking errors/time on task (Van Orden, Jung and Makeig, 2000: 226) 173 Process involved in applying parking brake to hold the vehicle stationary 189 Cause and effect diagram for vehicle failing to remain stationary 190 Fault tree analysis for parked unattended vehicle failing to remain stationary 191 Parking practice reported by respondents 195 Reasons provided for parking practice 196 Comparison between groups of parking practice on a slope and reasons provided 197

fC

7.1

14.2 14.3

14.4 14.5

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

14.6

16.1

16.2 16.3 16.4 16.5 16.6

9781472414694_Dorn.indb 10

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM

List of Figures

9781472414694_Dorn.indb 11

197 199 202 203 210 210 213 213

fC

op

y

16.7 Comparison of reported parking practice in car parks between groups 16.8 Cambridge hospital car parks 16.9 A comparison of manually operated (HB) with electronic (EBP) parking brakes for vehicles left in gear 16.10 Parked vehicles registered 3 years and less fitted with EPB 17.1 Driving simulator with motion system 17.2 Speed profile of the test track 17.3 Means (SD) of maximum deceleration (left) and energy consumption and energy recuperation (right) in deceleration situations by ACC versions (n = 24) 17.4 Scatterplots for speed and energy consumption while decelerating from 70 to 50 km/h for the three ACC versions (n = 1) 17.5 Means (SD) of scores for strength of the vehicle movements and pleasure in all deceleration situations for the three ACC versions (n = 24) 17.6 Means (SD) for maximum acceleration (left), overall energy consumption and in the acceleration process (right) for the three ACC versions (n = 24) 17.7 Means (SD) strength of the vehicle movements and pleasure evaluations in all acceleration situations for the three ACC versions (n = 24) 17.8 Means (SD) for maximum deceleration when approaching a corner (left), minimum speed when cornering (middle) and maximum acceleration when exiting the corner (right) for the three ACC versions (n = 24) 17.9 Means (SD) of fuel consumption for approaching the corner (left), cornering (middle) and leaving the corner (right) for the second sharp right corner (n = 24) 17.10 Means (SD) for the strength of the deceleration and pleasure when approaching the four corners and cornering in the three ACC versions (n = 24) 17.11 Means (SD) for overall energy consumption (left) and energy recuperation (right) in the car following situations for the three ACC versions (n = 24) 18.1 Process of accident data acquisition in the Audi Accident Research Unit 18.2 The five step method of accident causation coding 18.3 Comparison of accident causes allocated by the AARU and GIDAS 18.4 Distribution of human causes for accidents allocated by the AARU (n = 519)

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

xi

214 215 215

216 217 218 218 225 227 227 228

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM


5/30/2013 10:19:50 AM

List of Tables

3.6 3.7 4.1 4.2 4.3 4.4 4.5

y

3.5

op

3.4

Mean scores for subjective workload and the number of collisions 20 Participant details 28 Speeds chosen for the first part of the simulator study 29 Correlations between feeling of risk and task difficulty and feeling of risk and crash probability 31 Regression analyses for task difficulty, feeling of risk and probability of loss of control for all three road environments 32 Significant findings from split plot ANOVAs for feeling of risk, task difficulty and probability of a crash 33 Significant findings from split plot ANOVAs for age differences in driven speed in the second part of the simulator study 34 Significant findings from split plot ANOVAs for differences in the accuracy of speed perceptions in the second part of the simulator study 35 Instructor competence profile and tests used for the Dutch driver training exam 45 Number of test versions, total number of items and sample size chosen 51 Cut-off scores and cut-off reliability for the three theoretical tests 52 Psychometric report for the Final Performance Assessment Lesson 55 Correlations between performance on theory tests and Performance Assessment Lesson 56 Risky driving manoeuvres included in the simulated driving 66 Situations pre-study 67 Implemented situations 69 Study design 93 Drivers wearing a seat belt by gender and age group 112 Secondary task engagement by seat belt usage, gender and estimated age 113 Correlations between the candidate’s driving self-efficacy and instructor’s ratings and predictions about the candidate’s future driving 131 Prediction of male traffic rule violations in the six month period 132 Prediction of female traffic rule violations in the six month period 132 Prediction of male errors while driving in the six month period 133 Number of work-related crashes before and after the interventions 143

fC

2.1 3.1 3.2 3.3

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5.1 5.2 5.3 7.1 9.1 9.2

11.1

11.2 11.3 11.4 12.1

9781472414694_Dorn.indb 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM


xiv

oo

fC

op

y

12.2 Fuel use (litres/per 100 kilometres) before and after the interventions 143 15.1 Mean and standard deviation of feedback characteristics 182 16.1 Results of freedom of information requests to police constabularies 188 16.2 Initial survey results of reported parking practice 194 16.3 Comparison of parking practice in car parks between advanced motorists and health care groups 198 16.4 Results of parking practice observations in car parks 201 16.5 Vehicles observed in car parks registered 2009–2012 202 17.1 Parameters for the three ACC versions in the four different driving situations 211 17.2 Factors increasing energy efficiency and driving pleasure in the present electric vehicle 220 19.1 Frequency of self-reported crashes in previous three years and over lifetime at Time 1 238 19.2 Discrepancies in self-reported crash details at Time 2 239 20.1 Overview of the samples and scales used 250 20.2 Descriptive results for the samples, first wave 251 20.3 The Pearson correlations and partial correlations (controlled first for mileage only, and thereafter for mileage and years of driving) between the questionnaire scales and self-reported collisions since licensing and current number of penalty points on licence in the YDS sample, first wave (n = 9965) 252 20.4 The Pearson correlations and partial correlations (controlled for mileage) between the questionnaire scales and self-reported collisions since licensing and current number of penalty points on licence in the YDS sample, third wave (n = 1186) 252 20.5 The Pearson correlations and partial correlations (controlled for mileage) between the questionnaire scales and self-reported collisions since licensing and current number of penalty points on licence in the YDS Control sample, first wave. The change in amount of explained variance calculated as the difference in squared correlations divided by the squared zero-order correlation. When a correlation changed sign, no change was calculated (n = 1231) 253 20.6 The Pearson correlations and partial correlations (controlled for mileage) between the questionnaire scales and self-reported collisions since licensing and current number of penalty points on licence in the YDS Control sample, second wave. The change in amount of explained variance calculated as the difference in squared correlations divided by the squared zero-order correlation (n = 234) 253 20.7 The Pearson correlations and partial correlations (controlled for mileage) between the questionnaire scales and self-reported collisions for the last three years and current number of penalty points on licence

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM

List of Tables

20.8

21.1 21.2 21.3 21.4

fC

20.10

op

y

20.9

in the SS sample, first wave. The change in amount of explained variance calculated as the difference in squared correlations divided by the squared zero-order correlation (n = 8013) 254 The Pearson correlations and partial correlations (controlled for mileage) between the questionnaire scales and self-reported collisions for three years and current number of penalty points on licence in the SS sample, second wave. The change in amount of explained variance calculated as the difference in squared correlations divided by the squared zero-order correlation. When a correlation changed sign, no change was calculated (n = 407) 254 The Pearson correlations and partial correlations (controlled for mileage) between the questionnaire scales and self-reported collisions for three years and current number of penalty points on licence in the RLS sample, first wave. The change in amount of explained variance calculated as the difference in squared correlations divided by the squared zero-order correlation (n = 4807) 255 The Pearson correlations and partial correlations (controlled for mileage) between the questionnaire scales and self-reported collisions for three years and current number of penalty points on licence in the RLS sample, second wave. The change in amount of explained variance calculated as the difference in squared correlations divided by the squared zero-order correlation. When a correlation changed sign, no change was calculated (n = 961) 255 Mean scores on the DBQ factors 265 Factor structure of the modified DBQ 266 Pearson correlations between the variables 267 Logistic regressions with self-reported crashes and traffic offences in previous three years as dependent variable with Demographic figures at step one and DBQ figures at step two 269 Driving characteristics as a percentage of the sample 280 Predictors of the loss of points from the licence 283 Mean scores of the sample 284 Factor structure of the Polish Driving Cognitions Questionnaire (n = 211) 296 DCQ item mean scores and item-total correlations (n = 211) 297 Correlations between the Polish DCQ factors and the other continuous variables 297 Correlations between the DCQ and its subscales as well as measures of validity 298

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

xv

22.1 22.2 22.3 23.1

23.2 23.3

23.4

9781472414694_Dorn.indb 15

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM


5/30/2013 10:19:50 AM

Preface

fC

op

y

2013 marks the tenth anniversary of the introduction of the International Conference in Driver Behaviour and Training (ICDBT). Ten years ago, relatively little was known about how educational interventions might reduce crash risk and some evaluations suggested there may even be a negative outcome. It was my hope that the ICDBT would be a catalyst for encouraging research in this field to investigate how to design and deliver training to influence driver behaviour in a positive direction – especially since the value of education had been demonstrated amongst many other health-related behaviours. When scoping out the conference, I could not have imagined how well the event would be accepted by the academic and practitioner community with participation from road safety researchers from almost a hundred countries. Nor could I have anticipated that hundreds of academic papers would be presented and that Volumes 1–5 of the proceedings would be used as a source of literature by thousands of people. I am delighted that we have now published the sixth volume to mark the continuing success of the ICDBT. This volume represents about a quarter of the papers disseminated at the ICDBT6 with all abstracts published separately in association with the Institute of Ergonomics and Human Factors. In the 10 years since the ICDBT’s inception, we are in a different place. Since then, there has been a massive upsurge of publications and research activity in driver behaviour and training. Unfortunately, at the same time as the substantial increase in literature on road safety, there has been a catastrophic failure of governments to use this evidence base in the implementation of policies and legislation to reduce the numbers of people being killed and injured. As our knowledge increases, so does the global death rate from road traffic collisions, predicted to rise to 2.6 million people killed every year by 2030 if governments fail to take action. With regards to recent findings however, we are starting to see some promising results. For example, some research suggests that hazard perception training improves visual search strategies and reduces risk amongst both experienced and novice drivers, but there is little evidence that advanced vehicle handling skills in critical situations are beneficial for safety unless the driver is able to reflect on their own limited abilities in such circumstances. What the evidence also points towards is the importance of ensuring educational interventions are delivered according to the risk profiles of particular road users groups. Interventions which take account of specific behavioural risks and inform the structure and content of education is the main approach for the Cranfield University company DriverMetrics, with over 100,000 drivers taking part to date. Crucially, the method by which the intervention is delivered seems to have a positive effect on driver behaviour. There is increasing recognition that education

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM


xviii

oo

fC

op

y

focusing on developing the motivation to maintain good safety margins, dealing with time pressure at work, peer pressure etc. is far more important for managing the human factors in driving rather than knowledge about the rules of the road or developing advanced vehicle handling skills. Active participation using the group discussion method appears to be a key factor in encouraging self-reflection on these kinds of risks, alongside the development of an action plan with the solutions and strategies on how to reduce the risks identified. What still seems to be lacking however, are evaluations based on scientific principles to investigate educational effectiveness and behavioural changes amongst participants, particularly over the longer term. It has also been recognised that for an intervention to be effective, countermeasures are best served by on-going feedback to keep the driver’s behaviour on track. For example, driver education using e-learning over a period of time appears to extend the value of an initial intervention and lead to a significant change in behaviour compared with a one-off workshop. However, it is also clear that driver behaviour tends to drift back to its original form in most cases. One way of addressing this is to implement interventions that have the capacity to provide continuous feedback. Providing continuous feedback to drivers can be achieved with the introduction of sophisticated technology installed in the vehicle itself, providing the driver is not distracted. A notable innovation being offered by insurance companies in recent times has been the ability to measure g force and GPS position using black boxes and smart phone apps measuring speed, excessive braking, journey time and risk exposure. This technology can be used as an educational intervention to provide regular bespoke feedback after each journey and flag up personal strengths and weaknesses with safer behaviour being motivated by lower insurance premiums. In 2002 I was quoted as saying at a major conference that telematics would be the main way in which driver education will be delivered in the future. The British Insurance Brokers’ Association predicts that 500,000 drivers in the UK alone will have telematics in their vehicle within the next 12 months. So what will the next ten years bring? I believe the commercial sector will play a key role in improving road safety as organisations begin to recognise the human and financial costs of managing risk. Interventions will become increasingly technological and installed in the vehicle itself, with behavioural feedback being largely automated and online and controlled via a system of reward and punishment by organisations interested in reducing the human and financial cost of crashes. The challenge is to ensure that these in-vehicle systems assist in safer driving, do not distract the driver or produce unwanted behavioural outcomes. I would now like to extend my thanks to many organisations, individuals, contributors, sponsors and delegates without whose support the role of Conference Chair would not be possible. First, my thanks to Dr Mark Sullman for coorganising the event and co-editing these proceedings so that the task was much less daunting than it had been on previous occasions. I am grateful to Professor Heikki Summala and his group, who graciously assisted in the organisation of the

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 18

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM

Preface

ICDBT6 and its hosting at Helsinki University in August 2013. We would also like to acknowledge the support of two key professional associations, the IEHF and the International Association of Applied Psychologists (IAAP) – Division 13 Traffic and Transportation Psychology for their support. Lastly, I would also like to thank the following sponsors for kindly supporting the ICDBT 2013: Unilever, AA Drive Tech, Mercedes Benz Driving Academy and Shell. Keynote speakers were drawn from major authorities in the field led by Professor Heikki Summala. Professor Summala is Chair of Traffic Psychology and head of the Traffic Psychology Unit at Helsinki University and an accomplished luminary in driver behaviour whose work I have always personally admired. Professor Summala’s keynote was followed by leading researcher, Professor Nils Petter Gregersen, the Senior Research Director of the Swedish National Road and Transport Research Institute. Professor Gregersen has devoted his career to improving road safety and road safety education, particularly for vulnerable drivers. His study, employing a randomised controlled trial, which investigated the effect of driver education on crash involvement, is still the most often cited seminal paper in the field. Our keynote addresses closed with an excellent research overview on visual strategies and driver training by world-renowned experts in visual search, Associate Professors Peter Chapman and David Crundall of Nottingham University, UK. The keynote speakers were followed by a high calibre of contributors from many academic institutions and road safety groups delivering more than 80 high-quality presentations covering a range of topics in driver behaviour. I am grateful to the scientific committee, contributors, our sponsors, exhibitors and the delegates for their support. 2013 is also the 20th anniversary of the tragic death of my very dear friend and driver behaviour researcher at Aston University, Dr Tom Hoyes who was killed as a passenger in a road traffic incident. In your name, the work continues, Tom.

9781472414694_Dorn.indb 19

y

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 Lisa Dorn 29 Conference Chair 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

op

fC

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

xix

5/30/2013 10:19:50 AM


5/30/2013 10:19:50 AM

fC

op

y

Part 1 Driver Education: The Role of Experience and Instruction

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM


5/30/2013 10:19:50 AM

Chapter 1

Anticipation, Neural Function and Mastering Driving Timo Järvilehto, Veli-Matti Nurkkala, Kyösti Koskela and Jonna Kalermo Kajaani University of Applied Sciences, Finland

y

Introduction

fC

op

Anticipation is infrequently ascribed a special role in scientific explanations of behaviour, in spite of its abundance in everyday life. Anticipation may be seen when we approach a closed door and have the keys ready in the hand; when we come to a meeting and think about the presentation we are going to give; or when we open a book for reading and wonder what the writer might have to say. Also, in the number of synonyms of anticipation which are commonly used in everyday life, adjectives such as ready, proactive, or prospective; substantives such as readiness, preparation, expectation, or attitude; and verbs such as foresee, predict, or forestall. These words refer to events or processes preceding the actual acts to be performed, events that are probably quite significant when fast and skilled action is needed. When teaching driving in a driving school, the significance of anticipation is usually stressed, but it is not always clear what precisely the teacher meant by this concept. This is understandable, because in the present scientific study of human behaviour neither cognitive psychology nor neuroscience has been particularly interested in the concept of anticipation. Such neglect is understandable within the framework of the mainstream stimulus processing model as it is assumed that the most interesting and significant processes in the nervous system start with the appearance or presentation of the stimulus, the behaviour being essentially a result of information processing in the brain and nervous system. Even if the concept of anticipation is used, it is related to the process of waiting for the presentation of the stimulus, either as the activation of an inner model with which the stimulus may be compared (e.g., Rosen, 1985), or as a more or less general process assumed to facilitate the advance of processing the future stimuli. However, when skilled and fast action is considered there are several problems with the information processing model. If anticipation means only that there is an inner model waiting for the processing of environmental events, how is fast action possible? Furthermore, if stimuli must always be processed before actions can be carried out there is always a time lag between the environmental events

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 3

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM


4

op

y

and movements of the subject. How then could synchronisation with rapid environmental events be possible? It is the main thesis of the present chapter that anticipation is not a factor related to the expectation of the stimuli or modifying their processing, but it is the main principle of the organisation of the nervous system and the brain, which determines the outline of the prospective acts and the features of the environment that can be joined to the process and lead to the desired behavioural outcomes. In the state of anticipation the nervous system is not waiting for stimuli, but it is actively organising, with all necessary bodily and environmental constituents, to achieve the desired results of action. This preceding organisation will determine which significant environmental constituents (e.g., traffic lights, signs) can be selected for the subject’s realisation of the actions. Thus, anticipation means that before the execution of the components of the driving process (e.g., steering and using brakes), a system is created for extracting relevant constituents from the environment, and the system for realising the movements is configured in advance in such a way that synchronous action is possible at the moment when a critical environmental incident is present.

fC

Development of the Anticipatory System in Mastering Driving

oo

How then are such anticipatory systems formed? In fact, anticipatory systems start to develop during phylogenesis. The structure of the new-born organism anticipates the features of the environment that can be used in the maintenance of its life process. Thus, every organism anticipates something about its environment in the sense that it has a structure into which only certain parts of the environment may be fitted in order to achieve useful results. In the ontogenesis, anticipatory systems are created during learning and training that lead to the environmental constituents that may be selected directly during the execution of the task. It is this process that makes fast and skilled action possible. Let’s illustrate this process with an example. In the beginning we have the environmental events A, B, C, and D, and the corresponding neural processes a, b, c, and d that may be considered as separate reactions to the respective environmental events, and which occur with a delay after each event. The result of the whole process appears with some delay after D. When the sequence is repeated during training the neural events a, b and c get connected so that when A occurs the process proceeds from a to d in such a way that the neural events may be simultaneous or even precede the corresponding environmental events, and the final result of the whole process may appear simultaneously with, or even preceding, the final event D. Thus, learning to drive does not mean there is a simple acceleration of the driver’s reactions to the critical stimuli, but a complete reorganisation of the neural processes responsible for the driving actions. This is illustrated in Figure 1.1. In the beginning, when learning to drive, each component of the action is trained

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 4

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:50 AM

op

Change in the structure of action during training

fC

Figure 1.1

separately and the action consists of several results (e.g., pressing a pedal – R1 and turning the wheel – R2) leading to the final accomplishment of the task (e.g., turning in the corner – Final Result). During training neural systems are formed that correspond to different possibilities in accomplishing the task (alternatives), depending on the environmental possibilities. When the subject masters the task, these alternatives are activated simultaneously at the beginning of the task (i.e., the driver has potential neural systems activated simultaneously for pressing pedals and turning the wheel) and different environmental markers, they select from these potential alternatives the actions which are most suitable. Thus, during skilled action the subject is not making any decisions in respect to the different parts of the performance, but is simply carrying out one of the many potential alternatives, while all others are blocked due to the unsuitable environmental events.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5

y

Anticipation, Neural Function and Mastering Driving

A Neural Model for Anticipation

How is anticipation realised by the nervous system? In contrast to the neural processing model (stimulus-response model), the theory of the organismenvironment system (Järvilehto, 1998a) starts with the conception that the nervous system is not a system for responding to stimuli in the environment, but it is a system organised together with bodily elements and environmental constituents for anticipated behavioural results.

9781472414694_Dorn.indb 5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM


6

oo

fC

op

y

According to the theory (for details, see Järvilehto, 1998b), a neuron is not an element for information processing, but it is a living cell that has to maintain its metabolism. When the metabolism of the neuron is disturbed the neuron starts firing which is its method to restore its chemical conditions by influencing other neurons. Thus, starting with a single neuron, the activity of the neuron is not a response to a stimulus, but its firing is rather a sign of a disturbance in its metabolism. By influencing other neurons and joining neural networks eventually the muscles are influenced and are joined to the environmental constituents. The neuron may then restore its state to the previous condition, which leads to the cessation of its firing. If the activity of the neuron does not lead to a useful result, it simply dies. Actually, this is happening all the time in the nervous system. Thus, already at this level the activity of the neuron is not a reaction to a stimulus, but its activity anticipates a future result (i.e., the restoration of its metabolic conditions). Within the framework of this theory, an action potential of a neuron is not an information transmitter, but a disturbance at its membrane, which is the way to influence other neurons. Thus, the transmitters, for example, do not convey any information from one neuron to another over the synapse, as is commonly thought, but they are simply chemicals which may distort the metabolism of other neurons (excitatory synapses) or supply them with useful metabolites (inhibitory synapses). Furthermore, according to the organism-environment theory, sensory receptors or receptor matrices (different senses) are not literally receptive, but rather they create a direct connection to environmental constituents, which supports the formation of the whole organism-environment system. In this process efferent influences on receptors (see Järvilehto, 1999 for details) are of particular importance, because they condition the receptors for the selective use of environmental constituents that are needed to result in action. Similarly, the muscles are not only efferent organs, but they also contain afferents that have a special significance in the interplay with the receptors. Thus, both the muscles and the receptors have a similar innervation (afferent and efferent) and they act together in defining those parts of the environment that can be used in action achievement. In conclusion, the present considerations mean that anticipation is not just a special factor for making information processing more rapid or for preparing the muscles of the organism for quicker reactions to external stimuli. Anticipatory systems are traditionally considered to contain a model of the organisms’ goal, while according to the organism-environment system theory: anticipation is inherent in all living systems. According to the theory of the organism-environment system, anticipation is inevitable, because it follows from the structure of the system. Anticipation is based on the general organisation of the living systems, and it became especially effective with the advent of the nervous system.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 6

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM


Is Driving a Motor or a Sensory Process?

fC

op

y

Skilled behaviour, such as driving, is traditionally divided into motor and sensory components. However, on the basis of the theory of the organism-environment system, there is no motor activity in driving, as contrasted with sensory activity. Driving is a process that always involves the whole organism-environment system, leading to specific behavioural outcomes. From the psychological point of view, there is nothing motor in the motor cortex, as there is not anything sensory in the sensory cortices. The terms motor and sensory are anatomical, not functional concepts. The respective units in the brain do not carry out psychological operations; the neurons are only parts of a larger system in which psychological operations are accomplished. During driving there are, of course, changes in the functioning of neural units, but these changes are related to the achievement of new results, not to separate psychological functions. In this process both motor and sensory components are always necessary. When driving, the perception of traffic signs, for example, is not a linear process which starts at the sign located on the side of the road to its perception, but rather a circle involving both the sensory and motor organs (e.g., head and eye movements) as well as the events in the environment (see Järvilehto 1998a). A perceptual process does not start with the stimulus, but the stimulus is rather an end of this process. It is the anticipatory process that determines the environmental constituents that can be used as stimuli. The stimulus is like the last piece in a jig-saw puzzle. The last piece of the puzzle fits into its place only because all other pieces of the puzzle have been placed in a particular way. It is just this joining of the other pieces, their coordinated organisation that leaves a certain kind of hole into which the last piece may be inserted. Thus, it is just the preceding organisation of the other pieces which defines a possible last piece with which the puzzle may be finished. Exactly in the same way a stimulus is present only if there is an anticipatory organisation into which the stimulus may be fitted. Therefore, the stimulus is as little in a causal relation to the percept as the last piece of the puzzle is to the constructed picture (Järvilehto 1998a). Thus, the event appearing after the stimulus (i.e., the reaction) in the brain or in the behaviour (e.g., braking with the red traffic light) is the result of the anticipatory organisation preceding the stimulus. It is not a reaction to the stimulus, but rather a transition from one act to another which is made possible by the anticipatory organisation of the system, and only triggered by the stimulus. Thus, braking for a red traffic light, for example, reflects more about driver knowledge (and acceptance) of the traffic rules and driver experience of the relevant traffic situation than any simple driving skill. Is learning to drive a motor or a sensory process then? One could stress its motor aspects, because of the significance of steering and control of pedals, as well as eye movements when monitoring the events in the car and in the environment. However, one could also stress the perceptual point of view, because without

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

7

9781472414694_Dorn.indb 7

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM


8

oo

fC

op

y

vision it would be difficult to control the vehicle properly in the traffic. From the present point of view, neither description is accurate. If learning is defined as the differentiation and widening of the organism-environment system (Järvilehto, 2000), then it is clear that learning cannot be sensory or motor only, as the process of learning always involves many systemic constituents. Learning to master driving a vehicle consists essentially of the development of the prospective organisation of the organism-environment system for skilled sequences of actions, in which sensory and motor components are integrated. This process is not related to movements or perceptions separately, but rather to the formation of sensory-motor integration in the form of action systems for specific results, with this being the target of the training process. Such systems are formed when the trainee performs in real traffic, but such systems may also be partially formed in a simulated driving situation. In the latter case we may speak about mental training. This kind of training is not at all more mental than the former one, but it is related to the formation of the parts of the action systems necessary in the final accomplishment of the objectives of the training in the real environment. It is also important to stress that from the point of view of the organismenvironment theory, human learning has essentially a social character and presupposes the existence of consciousness (see Järvilehto, 2000). According to the organism-environment theory, consciousness develops in cooperation with other people, human learning being a process exceeding the borders of the individual organism-environment system. It is this larger organisation in which human learning is realised, and therefore all efficient learning presupposes the participation of the trainer and the trainee as well as all other supporting people. From this it follows that the task of the trainer in teaching driving is not that of information transmission (teaching in the classical sense), but rather the creation of the cooperative organisation in which the learning resources of the trainee may be realised. An essential characteristic of this kind of process is the trainee’s developmental opportunities, and discovering their personal style in the process of achieving the desired results, in cooperation with the trainer. Teaching driving is not a process of transmission; it is rather a process of organising the pre-existing skills of the trainee into a larger organisation, which in the beginning consists of the trainer (and all other relevant people) and the trainee, but with continuing training this becomes more differentiated and narrow until the trainee develops personal skills and is able to achieve the desired results without the immediate support of others. In conclusion, developing mastery of sensory-motor skill, such as driving, is not a process of motor learning going on in the motor cortex of the trainee or a process in the visual areas, but rather a deeply social process which is directed towards the creation of an integrative organisation which consists of many parts and participants. The brain is, of course, also an important component in such an organisation. However, the learning process is not confined to the brain only, but it also presupposes many other necessary components, such as the body, environmental possibilities and social interactions.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM


The Anticipatory System

fC

op

y

Driving is often conceptualised as a complex system of behavioural adaptation (Summala, 2007). Within the framework of the stimulus-response model, behavioural adaptation was defined by Summala (1996) in the following way “the driver is inclined to react to changes in the traffic system, whether they be in the vehicle, in the road environment, in road and weather conditions, or in his/her own skills or states, and that reaction occurs in accordance with his/her motives” (p. 189). From the point of view of the organism-environment system theory, such a conception of driving arouses several questions. Does the driver really react? How could one react to their own states? What exactly is a motive? Is it causally related to behaviour? What then does in accordance with motives mean? From a systemic point of view behavioural adaptation means configuring the system so that it produces the desired results. A desired result, for example, would be the safe completion of a trip, and the factors to which the driver is supposed to react are constituents of the system (e.g., changes in the vehicle or traffic environment). The driver does not react, but the organisation of the organismenvironment system anticipates certain configurations of the factors listed above. If some of the factors do not fit into the organisation, problems arise for the driver, and the driving system is reorganised in order to facilitate the achievement of the desired results. In order to illustrate this conceptualisation in detail, let’s look at the process of getting home from work (Result). What factors in the formation of the organismenvironment system are important in making this result possible? Firstly, we must look at the history of the system (i.e., at the organisation of the system that leads to the situation that the driver wants, which in this case is the motive to go home). Wanting (or the motive) is based on a certain kind of preceding history (e.g., the working day is over and their partner has asked them to come home). Thus, we have the start of the anticipatory system for achieving the expected result. Now, the driving system is created by fitting the different factors together. There is no reaction by the driver, but rather an anticipatory process in which the different factors are together used to achieve the result. If some factors do not fit together, the system must be reorganised. For example, if the car does not start, help must be sought. If the weather conditions are very bad, more time is needed for achieving the result, which may also change the driver’s plans. The system formation quality gets its expression in the emotions of the driver. Deviance from the anticipated conditions (e.g., the car does not start) may be the basis for negative emotions (e.g., anger) and lead to the search for new options. The emotions of the driver are not separate from their actions or in causal relation to the action, but they are an expression of the state of the driving system (see Järvilehto, 2001). If the sub-results are achieved in an organised manner there is a flow of action. In other words, transitions from one act to another are smooth, which is consciously experienced as comfort or generally as a positive emotion (e.g., joy).

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9

9781472414694_Dorn.indb 9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM


10

oo

fC

op

y

When looking at the vehicle in traffic we may use the zones described by Hall (1966): personal, social, and public. According to the organism-environment theory, these zones consist of objects that make different kinds of results possible: personal zone – caring of your own body; social – contact with close people, or the manipulation of objects; public – contact with strangers and the manipulation of distant objects. Certain results are anticipated with objects in each zone; if an object intrudes into the wrong zone, reorganisation of the system happens. The optimal flow of action presupposes that the objects with certain characteristics stay in the correct zone. If a tree suddenly appears in the personal zone during a drive, there is the danger of getting hurt (i.e., time to collision is too short). The anticipatory system is continuously estimating the possibility of intrusion of any given object into the wrong zone whilst driving (i.e., the system is well equipped with an accurate estimation of the time of the possible intrusion over the boundary of any given zone) (Summala, 2007). This also means that the system continuously determines the object’s properties on the basis of its organisation in the environment. The system must continuously assimilate environmental opportunities that make the flow of action possible. If an event (e.g., an object in the wrong zone) suddenly occurs that does not fit into the anticipatory organisation, the system must be reorganised and if this takes too much time a collision occurs. A comfortable and efficient driving situation consists of the optimal use of the environmental constituents in the process of achieving results. In other words, the continuously changing aspects of the driving environment must fit the anticipatory organisation of the organism-environment system. These environmental constituents may include such things as: a straight road, a curve, obstacles on the road, other vehicles going in the same or opposite directions, speed of vehicles, landscape (looking around), other people in the car, and pedestrians. The anticipatory system must be able to assimilate such factors; otherwise disorganisation of the system occurs that may lead to a crash. Conclusion

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

It is common sense that we prepare for many of the acts we are going to perform. It is, however, not so clear what happens during such preparation or in the anticipation of important events. In mainstream cognitive science, anticipation is seen as a process that makes the processing of important stimuli more effective and faster. However, the present considerations indicate that anticipation is not just a special factor for making information processing more rapid or for preparing the muscles of the organism to react faster to external stimuli. Anticipatory systems are traditionally considered as systems that contain the model of the result, whereas according to the organism-environment system theory, anticipation is immanent in all systems, the existence of which is dependent upon the results to be achieved. In the framework of the organism-environment system theory, anticipation is not an empirical fact, but it is one of the cornerstones of the whole theory.

9781472414694_Dorn.indb 10

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM


The present considerations indicate that in driver training it is not essential to try to carry out the required movements quicker and quicker, but one should concentrate on creating the anticipatory organisation and the action alternatives under varying circumstances. Here a driving simulator might be very useful, especially for training learners to master dangerous situations (e.g., bad weather conditions). Furthermore, as the constituents of the anticipatory organisation are created in relation to the final result to be achieved, the training should not be directed towards the mastery of the separate components as such, but it should always be related to the whole act, as would be present under normal conditions.

y

References

fC

op

Hall, E.T. (1966). The hidden dimension. New York: Doubleday. Järvilehto, T. (1998a). The theory of the organism-environment system: I. Description of the theory. Integrative Physiological and Behavioral Science, 33, 321–34. Järvilehto, T. (1998b). The theory of the organism-environment system: II. Significance of nervous activity in the organism-environment system. Integrative Physiological and Behavioral Science, 33, 335–43. Järvilehto, T. (1999). The theory of the organism-environment system: III. Role of efferent influences on receptors in the formation of knowledge. Integrative Physiological and Behavioral Science, 34, 90–100. Järvilehto, T. (2000). The theory of the organism-environment system: IV. The problem of mental activity and consciousness. Integrative Physiological and Behavioral Science, 35, 35–57. Järvilehto, T. (2001). Feeling as knowing – Part 2. Emotion, consciousness, and brain activity. Consciousness & Emotion, 2, 75–102. Rosen, R. (1985). Anticipatory Systems. Oxford: Pergamon. Summala, H. (1996). Accident risk and driver behaviour. Safety Science, 22, 103– 17. Summala, H. (2007). Towards understanding motivational and emotional factors in driver behaviour: Comfort through satisficing. In P.C. Cacciabue (ed.), Modelling driver behaviour in automotive environments: Critical issues in driver interactions with intelligent transport systems (pp. 189–207). Berlin: Springer.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

11

9781472414694_Dorn.indb 11

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM


5/30/2013 10:19:52 AM

Chapter 2

Does Driving Experience Delay Overload Threshold as a Function of Situation Complexity? Julie Paxion*, **, Catherine Berthelon* and Edith Galy**

Introduction

y

French Institute of Science and Technology for Transport (IFSTTAR), France; ** Aix-Marseille University (AMU), France

op

*

fC

Epidemiological studies show that young novice drivers have a risk of crash involvement two to four times higher than young experienced drivers (Triggs, 2004). Crash rates, which are very high in the first months, decrease rapidly after a few months experience (Mayhew, Simpson and Pak, 2003) and a few kilometres of driving (McKnight, 2006; Preusser, 2006; cited in Mayhew, 2007), and continue to decrease as driving experience increases (Williams, 2003). On the one hand, Rasmussen’s Skill Rule Knowledge (SRK) model (1987) demonstrates that driving skills are acquired with experience in three stages. Knowledge-based behaviours are controlled actions (slow and effortful) adopted by novice drivers who refer to their knowledge about the Highway Code and previous experiences. Skill-based behaviours are automatic actions (fast and effortless) which are adopted by experienced drivers (e.g., changing gear). Ruledbased behaviours are an intermediate step, which may be adopted by novice or experienced drivers. These are controlled actions which follow prescribed rules (e.g., stopping at a red traffic light). Considering this model, novice drivers often have a lack of routine automation (De Craen et al., 2008; Fuller, 2002), which can lead to driving impairments. On the other hand, situation complexity has an influence on the level of workload, as does the perception of the individual. Subjective workload is thus defined as the perceived cost, by an individual, of completing a task. If the activity is not entirely automated, performing the task implies making an effort. For complex tasks, the required effort can be too high for an individual’s capabilities and can thus result in overload, which is characterised by a level of workload where an individual’s performance is impaired. Despite subjective workload increases, driving performance can be maintained as compensatory mechanisms are gradually established with practice (Amalberti, 1996; Cegarra and Hoc, 2006). However, when subjective workload is either too high (overload) or too

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 13

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM


14

op

y

low (underload), depending on the links between the required tasks and a drivers’ 1 internal state (Hockey, 2003), driving performance will suffer (Meister, 1976; 2 3 cited in De Waard, 1996). Thus, for the same driving situation, the activity can be controlled or automated 4 depending on the individuals’ experience, with a higher effort required for novice 5 drivers than for experienced drivers (Patten et al., 2006). In other words, subjective 6 workload should increase with a lack of driving experience and with an increase in 7 situational complexity. Therefore, the threshold at which drivers report overload 8 not only depends on the complexity of the situation, but also on the skills acquired 9 10 during driving. Our main hypothesis is that the subjective overload threshold (i.e., the 11 subjective workload at which any increase results in a reduction in driving 12 performance) should be observed earlier for young novice drivers than for more 13 experienced drivers, especially in very complex situations. To test this hypothesis, 14 novice and experienced drivers were exposed to driving tasks with different levels 15 16 of complexity, while also completing questionnaires.

Participants

fC

Method

oo

Fifty-seven young drivers (33 males and 24 females) were divided into four groups, according to their driving experience. Two groups were composed of novice drivers who had obtained their driving licence within the last two months, with 15 Traditionally Trained Drivers (TTD)1 aged between 18–20 years old (M = 19, SD = 0.84) and 12 Early-Trained Drivers (AAC – Apprentissage Anticipé de la Conduite)2 aged 18 years. The two other groups were composed of 15 drivers aged 21 years old who were arriving at the End of their 3-year Probationary Period (EPP)3 and 15 Experienced Drivers (ED) who were aged between 23–30 years old (M = 27, SD = 2.97) with at least five years of driving experience.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Experimental setup

The experiment was carried out in the SIM²-IFSTTAR fixed-base driving simulator, which was equipped with an ARCHISIM object database (Espié, Gauriat and Duraz, 2005). The driving station comprised one quarter of a vehicle (see Figure 2.1). The image projection (30 Hz) surface filled an angular opening that spanned 1 TTD: 20 hours of driving lessons with an instructor. 2 AAC: 20 hours of driving lessons with an instructor and additional driving practice with an adult during 3,000 km., driving learning permitted to start at the age of 16. 3 EPP: from the driving licence exam, partial licence during three years with restrictions as speed limitation and only 6 points instead of 12.

9781472414694_Dorn.indb 14

17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:52 AM

Does Driving Experience Delay Overload Threshold

150° horizontally and 40° vertically. The vehicle had an automatic gearbox and was not equipped with rear view mirrors. Procedure

fC

op

y

Participants drove on three different rural driving situations (22.5 kms each) in a counterbalanced order. The simple and monotonous situation consisted of a straight national road with two way traffic, but without any traffic. The second situation was moderately complex and included both right and left hand corners (length = 600 m, radius = 300 m). The last and the most complex situation had double and sharper corners (length = 300 m, radius = 120 m), with oncoming traffic. In all three scenarios a pedestrian was also present. The pedestrians, were hidden by a billboard, a bus stop or a tree (in random order), and crossed the road around 2.7 seconds before the participant arrived at their location. Participants were instructed to drive at 90 km/h. The NASA-TLX questionnaire (Hart and Staveland, 1988) was used to assess the subjective level of workload after each session. The TLX is comprised of six dimensions: Mental Demands, Physical Demands, Temporal Demands, Performance, Effort and Frustration. For each dimension, participants estimated their workload during the last drive on a 20 point scale (0 = Very low to 20 = Very high). The questionnaire had been modified in order to investigate the subjective workload associated with the different parts of the three scenarios. In other words, participants were asked to rate the level of workload imposed by each scenario (the overall scenario) and each condition within each scenario (i.e., straight road, corners, traffic and pedestrians) using the six dimensions of the TLX.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

15

Figure 2.1

9781472414694_Dorn.indb 15

Driving simulator

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:53 AM


16

Statistical analysis Subjective level of workload and objective behaviour (number of collisions with pedestrians) were analysed. Polynomial regressions were carried out in order to test two models: • Model 1: The effect of situation complexity and driving experience on the subjective workload attributed to pedestrians, • Model 2: The effect of situation complexity, driving experience and workload attributed to pedestrians and the number of collisions with these pedestrians.

op

y

For all analyses, statistical significance was fixed at p < 0.05. Significant effects were further investigated using post hoc analyses for pairwise comparisons and simple linear regressions used to predict the dependent variables. Results

fC

Model 1: Effects of situation complexity and driving experience on subjective workload

oo

In this model, all predictors accounted for 12% of the variance in subjective workload. Subjective workload was significantly influenced by situation complexity (linear effect β = 0.15; p < 0.05). As expected, subjective workload increased as driving situations became more complex. In order, the means for the simple situation, through to the most complex situation, were 11.01, 12.12 and 12.44 (SDs = 3.96, 3.62 and 3.85, respectively). However, post hoc tests did not reveal any differences between each situation in terms of complexity.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Figure 2.2

9781472414694_Dorn.indb 16

Predictors of subjective workload

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:53 AM


Subjective workload decreased significantly with driving experience (linear effect β = −0.27; p < 0.001 and nonlinear effect β = −0.23; p < 0.01) (see Figure 2.3): Traditionally Trained Drivers (TTD) had higher scores than Experienced Drivers (ED). Early-Trained Drivers (AAC) had higher scores than drivers arriving at the End of their Probationary Period (EPP) and also had higher scores than Experience Drivers. Furthermore, scores for TTD were lower than those for AAC. No significant interaction between driving experience and situation complexity was observed on subjective workload (β = −0.43; ns).

fC

op

y

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Figure 2.3 Effects of driving experience on subjective workload 24 25 26 Model 2: Effects of situation complexity, driving experience and subjective 27 workload on the number of collisions with pedestrians 28 All predictors together accounted for 25% of the variance in the number of 29 collisions. A significant main effect was observed for subjective workload (linear 30 effect β = 1.05; p < 0.001). An increase in subjective workload was related to an 31 32 increase in the number of collisions. 33 34 35 36 37 38 39 40 41 42 43 44 Figure 2.4 Predictors of the number of collisions

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

17

9781472414694_Dorn.indb 17

5/30/2013 10:19:56 AM


op

Figure 2.5 Effects of subjective workload on the number of collisions

oo

fC

Situation complexity significantly increased the number of collisions with pedestrians (linear effect: β = 1.14; p < 0.01 and nonlinear effect: β = −0.29; p < 0.001). This number was lower in simple situation (M = 0.44, SD = 0.76) than in moderately (M = 0.82, SD = 0.66) and very complex situations (M = 0.88, SD = 1.00). A significant interaction effect between situation complexity and subjective workload (linear effect: β = −3.27; p < 0.001 and nonlinear effect: β = 2.11; p < 0.0001) indicated that collisions increased with the increase of subjective workload in simple (β = 0.47; p < 0.001) and very complex situations (β = 0.52; p < 0.0001), whereas in moderately complex situations, subjective workload had no effect on the number of collisions (β = −0.05; ns) (see Table 2.1). No main effect of driving experience (β = −0.63; ns) and no significant interaction effects between driving experience and situation complexity (β = 0.17; ns) were found, neither was there an interaction effect between driving experience and subjective workload (β = 1.38; ns) (see Table 2.1). It is important to note that the subjective workload attributed to pedestrians was not normally distributed for the traditionally trained or the early-trained drivers. A large dispersion between the participants of each group regarding the number of collisions was also observed. A rise of subjective workload attributed to pedestrians significantly increased the number of collisions for traditionally trained novices (β = 0.33; p < 0.05), earlytrained novices (β = 0.45; p < 0.01) and drivers with three years of experience (β = 0.31; p < 0.05) (see Table 2.1). No interactions between driving experience, situation complexity and subjective workload on the number of collisions (β = 0.01; ns) were observed. As mentioned previously, the large dispersion of the data could explain this result.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

y

18

9781472414694_Dorn.indb 18

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:57 AM

9781472414694_Dorn.indb 19

5/30/2013 10:19:58 AM

Figure 2.6 Effects of subjective workload on the number of collisions depending on situation complexity


20

Indeed, further analyses showed that the rise in subjective workload attributed to pedestrians significantly increased the number of collisions with pedestrians, but only in the most complex situations for novices with traditional learning (β = 0.54; p < 0.05) and early-trained novices (β = 0.68; p < 0.05) (see Table 2.1). Table 2.1

Mean scores for subjective workload and the number of collisions Subjective workload SD

M

SD

11.01 12.12 12.44

3.96 3.62 3.85

0.44 0.82 0.88

0.76 0.66 1.00

12.00 14.43 11.42 10.08

4.20 3.65 3.36 2.92

0.84 0.97 0.58 0.51

0.88 1.08 0.72 0.59

11.22 11.91 12.88

4.53 4.14 4.05

0.53 0.93 1.07

0.74 0.70 1.10

13.42 15.57 14.32

4.22 2.76 3.81

0.83 0.92 1.17

1.11 0.79 1.34

11.27 10.91 12.09

3.17 2.61 4.21

0.40 0.53 0.80

0.63 0.64 0.86

8.62 10.77 10.87

2.64 2.95 2.78

0.07 0.93 0.53

0.26 0.46 0.64

op

Situation complexity: Simple Moderately complex Very complex

Number of collisions

y

M

fC

Driving experience: Traditionally Trained Drivers (TTD) Early-Trained Drivers (AAC) Drivers at the End of the Probationary Period (EPP) Experienced Drivers (ED)

oo

Driving experience and situation complexity: TTS: Simple Moderately complex Very complex AAC: Simple Moderately complex Very complex EPP: Simple Moderately complex Very complex ED: Simple Moderately complex Very complex

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Discussion This driving simulator research aimed to identify whether driving experience delayed the point at which drivers reached their subjective overload threshold. The two regression models used here highlight the fact that situation complexity increased both subjective workload attributed to pedestrians and the number of collisions with these pedestrians. However, only a global effect on subjective

9781472414694_Dorn.indb 20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:59 AM


fC

op

y

workload (no difference between each situation) was found, whereas there were fewer collisions in simple situations compared to moderately and very complex situations. Therefore, complexity between the situations tested did not vary enough to produce different levels of subjective workload when confronted by unexpected pedestrians crossing. It could be, therefore, that even if moderately complex and very complex situations included corners, their presentation was too repetitive to modify subjective workload and thereby the number of collisions from the simple situation to the two complex ones. Moreover, independently of subjective workload, objective workload could have provoked the increase in the number of collisions. Indeed, human errors in traffic caused by objective mental workload are sometimes considered to be a substantial cause of traffic accidents (Smiley and Brookhuis, 1987; cited in Brookhuis and De Waard, 2010). An increase in driving experience did not influence the number of collisions, but it did increase the subjective workload attributed to pedestrians. Contrary to our hypothesis, traditionally trained drivers had lower subjective workload scores than Early-Trained Drivers, and had similar scores with drivers at the End of their Probationary Period. These results could be due to a high dispersion in the number of collisions in each of the four groups and of the subjective workload among Traditionally Trained Drivers and Early-Trained Drivers. Moreover, age could be an additional factor which could have influence these results, considering that Traditionally Trained Drivers were older (M = 19) than Early-Trained Drivers (M = 18). Contrary to our hypothesis, the interaction effect between situation complexity and driving experience neither increased the subjective workload attributed to pedestrians nor the number of collisions. The absence of subjective workload differences between all situations in the pairwise comparisons and the absence of a driving experience effect on the number of collisions could explain this result. As expected, an increase in subjective workload adversely affected driving performance through an increase in the number of collisions. However, there were no significant interaction effects between driving experience and subjective workload, nor between driving experience, situation complexity and subjective workload on the number of collisions. As seen previously, the high dispersion of the data and the age differences could explain the lack of interaction effects. However, the subjective overload threshold was reached for all groups, except for the most experienced drivers. Therefore, it seems that those with less than five years of driving experience relied on controlled knowledge-based or ruled-based behaviours, whatever the situation was, while the more experienced drivers had a skill-based behaviour with some automatic driving schemes leading to a decrease in subjective workload and allowed the appropriate manoeuvres. Considering the detailed results of each group in each situation, the subjective overload threshold was only reached by novice drivers (Traditionally Trained and Early-Trained) in the most complex situation. Therefore, the additional kilometres travelled by Early-Trained Drivers, compared with Traditionally Trained Drivers, is not enough to differentiate them in managing unexpected situations, such as a pedestrian

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

21

9781472414694_Dorn.indb 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:59 AM


22

suddenly crossing the road. Moreover, this result shows that the subjective overload threshold was not reached after three years of driving experience (EPP and ED groups), even when the situation was very complex. Drivers arriving at the End of their Probationary Period probably start to switch between automatic and controlled processing, and are thereby adopting ruled-based behaviour more efficiently than novice drivers. To sum up, this study reveals a progressive acquisition of automatic skills which gradually delays subjective overload threshold with learning. Limitations of the present study

op

y

Experimentation in simulators involves some biases, as drivers know that they are not in danger and they may adopt more risky behaviours than they would in reality. Conclusion

oo

fC

Training in a simulator with complex or/and unexpected situations may help young novice drivers to increase their overload threshold and thereby to be more prepared to adequately manage risky situations. The present study is only based on subjective workload, but physiological data (e.g., from an electrocardiogram) could reveal more precise results concerning overload threshold. It would therefore be interesting to compare subjective workload to physiological levels of workload. Further analyses are currently underway in order to identify the effects of other explanatory variables of overload, such as subjective levels of tension and vigilance (Conard and Matthews, 2008; Brookhuis, De Waard, Kraaij and Bekiaris, 2003). Acknowledgements

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

The authors thank the simulation team at IFSTTAR (LEPSIS: Laboratoire Exploitation, Perception, Simulateurs et simulations), notably Isabelle Aillerie for designing the displays. References Amalberti, R. (1996). La conduite des systèmes à risques, Paris: PUF. Brookhuis, K.A., and De Waard, D. (2010). Monitoring drivers’ mental workload in driving simulators using physiological measures. Accident Analysis and Prevention, 42, 898–903. Brookhuis, K.A., De Waard, D., Kraaij, J.H., and Bekiaris, E. (2003). How important is driver fatigue, and what can we do about it? In D. de Waard, K.A.

9781472414694_Dorn.indb 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:59 AM


fC

op

y

Brookhuis, S.M. Sommer, and W.B. Verwey (eds), Human factors in the age of virtual reality (pp. 191–207). Maastricht, Netherlands: Shaker Publishing. Cegarra, J., and Hoc, J.M. (2006). Cognitive styles as an explanation of experts’ individual differences: A case study in computer-assisted troubleshooting diagnosis. International Journal of Human-Computer Studies, 64(2), 123–36. Conard, M., and Matthews, R.A. (2008). Modeling the stress process: Personality trumps stressors in predicting strain. Personality and Individual Differences, 44(1), 171–81. De Craen, S., Twisk, D.A.M., Hagenzieker, M.P., Elffers, H., and Brookhuis, K.A. (2008). The development of a method to measure speed adaptation to traffic complexity: Identifying novice, unsafe, and overconfident drivers. Accident Analysis and Prevention, 4, 1524–30. De Waard, D. (1996). The measurement of drivers’ mental workload. Traffic Research Center, Thesis, 127 p. Espié, S., Gauriat, P., and Duraz, M. (2005, September). Driving simulator validation: The issue of transferability of results acquired on simulator. Paper presented at the Driving Simulation Conference North-America (DSC-NA 2005), Orlondo, FL. Fuller, R. (2002). Human factors and driving. Human Factors for Highway Engineers, 77–97. Hart, S.G., and Staveland, L.E. (1988). Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research. Advances in Psychology, 52, 139–83. Hockey, G.R.J. (2003). Operator Functional State as a Framework for the Assessment of Performance Degradation. In G.R.J. Hockey, A.W.K. Gaillard and O. Burov (eds), Operator functional state: The assessment and prediction of human performance degradation in complex tasks. Amsterdam: IOS Press. Mayhew, D.R. (2007). Driver education and graduated licensing in North America: Past, present, and future. Journal of Safety Research, 38(2), 229–35. Mayhew, D.R., Simpson, H.M., and Pak, A. (2003). Changes in collision rates among novice drivers during the first months of driving. Accident Analysis and Prevention, 35(5), 683–91. Patten, C.J.D., Kircher, A., Östlund, J., Nilsson, L., and Svenson, O. (2006). Driver experience and cognitive workload in different traffic environments. Accident Analysis and Prevention, 38(5), 887–94. Triggs, T.J. (2004). Simulation evaluation of driver performance changes during the early years of driving. Proceedings of the driving simulation conference, Paris, France, pp. 421–30. Rasmussen, J. (1987). The definition of human error and a taxonomy for technical system design. In J. Rasmussen, K. Duncan and J. Leplat (eds), New technology and human error (pp. 23–30). Chichester, UK: Wiley. Williams, A.F. (2003). Teenage drivers: Patterns of risk. Journal of Safety Research, 34, 5–15.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

23

9781472414694_Dorn.indb 23

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:59 AM


5/30/2013 10:19:59 AM

Chapter 3

Risk Allostasis: A Simulator Study of Age Effects Britta Lang*, Andrew M. Parkes* and Michael Gormley** *

TRL, UK; **Trinity College Dublin, Republic of Ireland,

y

Introduction

fC

op

Unsurprisingly, the concept of risk has been at the heart of driver behaviour research for many years. Road traffic crashes carry the risk of significant injury or loss of life; so understanding how risk affects driver decision making has important implications for the development of effective counter-measures. These could span education and training, modifications of the road infrastructure and the development of in-vehicle technologies aimed at either preventing the occurrence of crashes or minimising their negative impact on those involved. Early research into driver behaviour centred on the concept of accident proneness and aimed to develop differential models of accident involvement by attempting to identify stable traits, biological characteristics and upper performance limits that could reliably identify those drivers with an above average risk of being involved in a crash. However, the significant associations identified between these driver attributes and accident involvement were too small to be of practical or theoretical value (Haight, 1986). In addition to the limited success with identifying accident-prone drivers, two Scandinavian studies (Johannson and Rumar, 1966; Johansson and Backlund, 1972), which required drivers to recall traffic signs along a route, indicated that driving behaviour was not merely determined by the upper performance limits of the driver, but that motivation modulated drivers’ perceptual processes and decision making: drivers demonstrated significantly better recall of signs they regarded as important. Furthermore, the notion of driving as a self-paced task had emerged through an on-road study published by Taylor (1964) which investigated the Galvanic Skin Response rate (GSR) of drivers in different road environments as a measure of the subjective risk (anxiety level) experienced. Taylor interpreted his finding to mean that average skin response rates of drivers did not vary significantly across different environments as an indication that drivers are sensitive to changes in risk and adaptively vary their behaviour in response to that perception. Assuming a basic motivation to make progress he posited that drivers produced the level of risk they wished to take by adapting their driving speed accordingly.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 25

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:59 AM


26

oo

fC

op

y

This new interpretation of drivers as purposeful creators of the driving task in a dynamically changing environment led to the emergence of motivational models of driver behaviour in the eighties. These models address the potential conflict between the desire to progress versus that of maintaining safety and try to explain how drivers manage risk or task difficulty (Carsten, 2007). Representatives of this particular genre comprise the Theory of Risk Homeostasis (Wilde, 1982, 1988, 1989), the Threat Avoidance Model (Fuller, 1984), the Zero Risk Theory (Näätänen and Summala, 1974, 1976; Summala and Näätänen 1988, Summala 1996, 1998, 2000), and the Task-Capability Interface Model (Fuller 2000, 2005a, 2005b), with its associated Theory of Risk Allostasis (Fuller, 2009). The latter two models particularly have been the subject of recent experimental work and have sparked a lively debate in the research community. Näätänen and Summala (1974, 1976) asserted that driver decisions are not influenced by the perceived risk of a crash and therefore referred to their theory as the Zero-Risk Theory of driving. Later, in a different study, these same authors (Summala and Näätänen, 1988) emphasised the role of motivational influences on driver behaviour and of adaption processes which are attributed to exposure-related changes in driver perception and cognition. Driving decisions are assumed to be governed by the balancing of inhibitory motives (subjective risk) and excitatory motives. Such excitatory motives are additional to the desire to make progress and are posited to be influenced by personality, the driver’s state and journey-related motives, such as time gains or thrill seeking. Subjective risk in their model is defined as a feeling of uncertainty or anxiety located in a subjective risk monitor, which is only activated when a critical threshold of subjective risk is exceeded, typically through the violations of learned safety margins. When subjective risk exceeds that critical threshold and the risk monitor is activated, it can affect ongoing behaviour or future decision making with the aim of reducing subjective risk (typically by slowing or taking evasive action). Similarly, Fuller (2000; 2005b; 2009) argues that it is not the objective risk of a crash that determines driving behaviour, but a subjective feeling of risk. According to his Risk Allostasis Theory drivers aim to target and maintain a range of task difficulty and associated feelings of risk, predominantly by manipulating driving speed. This preferred range of difficulty is assumed to fluctuate, depending on internal and external conditions. The difficulty of the driving task in turn arises out of the dynamic interplay of the two main components of the Task Capability Interface model: task demands and driver capability. Simply put, if the capability of the driver exceeds the demands of the driving task (C > D), the driver progresses safely; if the demands of the task exceed available capability (D > C), and task difficulty is very high, the driver may lose control over the vehicle and (partly depending on other road users’ actions) a collision may occur. Driver capability, according to Fuller, is influenced by stable biological characteristics such as processing speed, by factors that slowly change over time such as age and experience, and by the comparatively more mercurial human factors such as fatigue, motivation and emotion.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 26

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:59 AM

Risk Allostasis: A Simulator Study of Age Effects

Experimental work testing the tenets of the Risk Allostasis Model to date, including Fuller’s own research (Fuller, McHugh, and Pender, 2008) and replications (Kinnear, Stradling, and McVey, 2008, Lewis-Evans and Rothengatter, 2009), has comprised two video-based and one simulator-based study in which participants were asked to rate short driving videos of incremental speed increases in different road environments. The analyses of drivers’ ratings of their feelings of risk, difficulty of the task and likelihood of a crash, in relation to the driving scenes shown, have consistently found significant positive correlations between speed and task difficulty. Furthermore, the available evidence supports the notion that feeling of risk tracks ratings of task difficulty more closely than the subjective estimate of the likelihood of a crash. However, whilst the associations are supported per se, there is still controversy whether feeling of risk increases linearly with speed and task difficulty (Risk Allostasis Model) or only after a certain threshold has been reached (Zero-Risk Theory). The results of the simulator study conducted by Lewis-Evans and Rothengatter (2009) point towards the existence of a threshold model, whilst linear increases in feelings of risk were observed by Fuller et al. (2008) and Kinnear et al. (2008). The experimental work to date has been limited with regard to the participant samples used, mostly drawing on available student populations. Only Kinnear et al. (2008) introduced driving experience as an independent variable and included learner drivers, inexperienced drivers (< 3 years driving experience) and experienced (> 3 years driving experience) in their study. The findings indicated that driving experience did not affect ratings of task difficulty and feeling of risk, but significantly reduced subjective estimates of crash likelihood. The present study expanded on the existing body of research by introducing age as an independent variable and investigated systematic differences between young, middle-aged and older drivers for the posited associations between task difficulty, feeling of risk and estimations of crash risk in a validated driving simulator. Whilst older drivers have attracted little interest from transport policy makers, due to their comparatively low accident involvement, the current and predicted demographic change towards much greater numbers of older people will inevitably change the age composition of the driving population. Understanding the perception of risk as a function of age can provide important information to underpin the design of traffic interventions that ultimately benefit all driver age groups.

fC

op

y

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 Method 38 39 Participants 40 Thirty healthy, current drivers from three age groups (young, middle-aged and 41 older drivers) were recruited from the TRL participant pool and took part in the 42 43 study. Table 3.1 provides an overview of the participant details. 44

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

27

9781472414694_Dorn.indb 27

5/30/2013 10:19:59 AM


28

Table 3.1

Participant details

Group

Young drivers

Middle aged drivers

Older drivers

Group age range

21–25 years

35–45 years

65+ years

Sex

5 females, 3 males

6 females, 5 males

6 females, 5 males

Mean driver age (years)

M = 23.4, SD = 1.8

M = 38.5, SD = 2.6

M = 67.9, SD = 2.6

Mean years since licensure

M = 6.0, SD = 1.8

M = 21.4, SD = 2.7

M = 48.0, SD = 5.6

Mean weekly mileage

M = 95.0, SD = 60.9 M = 191.8, SD = 177.4 M=129.5, SD = 85.5

y

Design

oo

fC

op

The study had two parts. The first part required each participant to complete 18 drives in the simulator at fixed speed (low, medium, high), per road environment (rural, urban, dual carriageway), with other road users either present or absent (ambient risk). The vehicle was accelerated automatically to the target speed (comparable to driving with cruise control) and the driver had simply to steer the vehicle for approximately 20 seconds, before the drive was stopped. The second part involved 12 drives at free speed choice. Here, the driver was in full control of the vehicle. In six drives participants were asked to drive at their preferred speed; in the other six drives, participants were asked to drive at the maximum speed they would choose if they were late for an important appointment. In both conditions, other road users were either present or absent in half of the drives (ambient risk). For both parts of the study the speedometer of the car was occluded to ensure that participants could not see the speed displayed. Equipment

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

The study was carried out in the TRL Driving Simulator (DigiCar). DigiCar consists of a medium sized family hatchback (Honda Civic), surrounded by four 3 x 4 metre projection screens, giving 210º front vision and 60º rear vision (see Figure 3.1). The road images were generated by four computers running SCANeR II software and were projected onto the screens by four Digital Light Processing (DLP) projectors at a resolution of 1280 × 1024 pixels, providing a screen resolution of approximately 13 pixels per inch. Images are refreshed at a rate of 60Hz whilst data is sampled at a rate of 20Hz. Electric motors supply motion with three degrees of freedom (heave, pitch and roll) whilst engine noise, external road noise, and the sounds of passing traffic are provided by a stereo sound system. Three sections of road were created for the current study, including an inner city area, a winding rural road and a straight dual carriageway. Two conditions per

9781472414694_Dorn.indb 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:59 AM


The TRL DigiCar (left) and instructor station (right)

y

Figure 3.1

fC

op

environment were created for each part of the trial, one in which no other road users were present in the scene and one where opposing smooth flowing traffic at appropriate speed was presented. For the first part of the trial which comprised driving at fixed speed, a high, medium and low speed condition was created for each of the three road environments. The speeds were chosen on the basis of the actual speed distributions on these road types in Great Britain in 2007 (Department for Transport, 2008), as shown in Table 3.2. The medium speed condition was chosen as the speed limit of the respective road type, the low and high speed condition were chosen as the speed two standard deviations below or above the actual mean driven speed on British roads of these particular types. For the rural road condition this led to three speeds that were not equidistant and associated distortions of the ratings. The rural road conditions were therefore excluded from further analysis.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

29

Table 3.2 Speed condition

Speeds chosen for the first part of the simulator study (fixed speed condition)

Explanation

Urban

Rural

Dual carriageway

Low

The speed that represented two standard deviations below the actual mean speed of driving on the particular road type in Great Britain

16 mph

28 mph

49 mph

Medium

The actual speed limit on that particular road type in Great Britain

30 mph

60 mph

70 mph

High

The speed that represented two standard deviations above the mean speed on the particular road type in Great Britain

44 mph

68 mph

89 mph

9781472414694_Dorn.indb 29

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:19:59 AM


30

Simplified Latin squares were used to permutate the order of the presentation of road environments and driving speeds across participants for the first trial part and across road environments and driving condition for the second part of the study, to avoid likely order effects of incremental speed condition presentation. Procedure

oo

fC

op

y

After briefing, participants completed a five minute drive to familiarise themselves with the car controls and dynamic vehicle reactions. During familiarisation, the speedometer was not occluded. After the car was started, it automatically accelerated to the target speed and then kept that speed constant using a cruise control system. After approximately 20 seconds, the experimenter stopped the drive and read out nine questions to the participants via a speaker system and recorded their responses. Participants were asked to estimate the speed at which they had just been driving. This was followed by seven rating questions, where participants had to judge on 7-point Likert scales (1 = Not at all to 7 = Extremely), how difficult, risky, stressful, dangerous, effortful and enjoyable the drive had felt and how nervous they had been. The ratings, in addition to task difficulty, of feeling of risk and crash probability were included to replicate and advance on similar rating dimensions included by Kinnear et al. (2009) and Lewis-Evans and Rothengatter (2009) and are not included in the present paper. The last question asked participants how often they thought they would have a crash if they drove that section of the road at this speed a hundred times. The experimenter recorded all participant responses before starting the next drive. The subsequent second part of the study comprised the completion of another 12 short drives in the simulator, four in each of the three road environments. Here participants were fully in control of the vehicle. They were instructed to either drive at their preferred speed or were asked to drive at the speed they would choose if they were late for a very important appointment. Once participants had reached their preferred or maximum speed in the second part of the trial, they announced this to the experimenter and maintained this speed for approximately 20 seconds, before the drive was stopped. After completion of each drive, participants were asked at which speed they believed they had just been driving, with the experimenter recording the response.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Results Risk, task difficulty and speed Participants’ average ratings of task difficulty and feeling of risk increased with speed for both urban roads and the dual carriageway; the increases were considerably steeper in the urban environment compared to those on the dual carriageway (Figure 3.2).

9781472414694_Dorn.indb 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:00 AM


op

y

Figure 3.2 Mean ratings and standard deviations for ratings of task difficulty, feeling of risk and likelihood of a crash on urban roads and on the dual carriageway

fC

Pearson Product Moment correlation coefficients were calculated to assess the relationship between ratings of task difficulty and feeling of risk in the two road environments for the three speed conditions (Table 3.3). The association was highly significant for all conditions; the strength of these correlations increased with speed for the urban road environment and decreased for the dual carriageway. In comparison, the correlations between feeling of risk and estimated likelihood of a crash were considerably lower and did not reach significance for the low and medium speed condition on the dual carriageway. Additionally, separate stepwise multiple regression analyses were conducted for both road environments with speed (low, average, high), ambient risk (present, absent) and age (young, middle-aged, old) as predictors and task difficulty, feeling of risk and subjective probability of a crash as dependent variables (Table 3.4).

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

31

Table 3.3

Correlations between feeling of risk and task difficulty and feeling of risk and crash probability Correlation between task difficulty and feeling of risk

Low speed

Average speed

High speed

Urban road

0.86**

0.82**

0.81**

Dual carriageway

0.74**

0.82**

0.87**

Correlation between feeling of risk and crash probability Low speed

Average speed

High speed

Urban road

0.59**

0.70**

0.45**

Dual carriageway

0.14

0.11

0.34**

(** p < 0.01)

9781472414694_Dorn.indb 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:00 AM


32

Table 3.4

Regression analyses for task difficulty, feeling of risk and 1 2 probability of loss of control for all three road environments

Task difficulty

Urban roads

Dual carriageway

r2

Beta

t

r2

Beta

t

Speed

0.37***

0.61

10.67

0.20***

0.55

6.68

Age

0.06***

0.25

4.3

Ambient risk

ns

ns 0.04**

0.20

3.15

0.25***

0.50

7.78

0.42***

0.65

11.37

Age

0.06***

0.24

4.45

Ambient risk

ns

Crash risk Speed

0.12***

0.35

Age

0.07***

0.27 ns

0.03**

0.19

2.92

4.90

0.06**

0.23

3.21

3.94

fC

Ambient risk

ns

op

Speed

y

Feeling of risk

ns ns

* p < 0.05, ** p < 0.01, *** p < 0.001

oo

Speed emerged as the most important predictor of task difficulty for both road environments. On urban roads it explained 37 per cent of the variance of the task difficulty ratings ( p < 0.001) versus 20 per cent on the dual carriageway ( p < 0.001). As a predictor of feeling of risk, speed explained 42 per cent of the variance in the ratings on urban roads ( p < 0.001) and 25 per cent of the variance on dual carriageways. Regression analyses indicated that speed also significantly predicted subjective probability of a crash on urban roads and dual carriageways; however, the association was considerably weaker, 12 per cent explained variance on residential roads ( p < 0.001) and 6 per cent on the dual carriageway ( p < 0.01), than that observed between speed and task difficulty and speed and feeling of risk.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Age effects on task difficulty, feeling of risk and probability of a crash To explore age effects in ratings of the feeling of risk, of task difficulty and estimates of crash probability obtained in the fixed speed part of the simulator, trial data were analysed using split plot ANOVAs with age group (young, middle aged, old) as between and speed (slow, average, high) and risk condition (ambient risk present, ambient risk absent) as within-subject factors. Separate sets of ANOVAs were carried out for the two road environments. The analyses found significant age

9781472414694_Dorn.indb 32

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:00 AM


Table 3.5

Urban road

Significant findings from split plot ANOVAs for feeling of risk, task difficulty and probability of a crash Feeling of risk

Task difficulty

Crash probability

Speed

F(2, 27) = 51.12 p < 0.001 partial η2 = 0.65

F(2, 27) = 44.27 p < 0.001 partial η2= 0.62

F(2, 27) = 11.51 p < 0.001 partial η2 = 0.30

Age

F(2, 27) = 4.75 p < 0.05 partial η2 = 0.26

F(2, 27) = 3.45 p < 0.05 partial η2= 0.20 F(1, 27) = 5.81 p < 0.05 partial η2 = 0.18

y

Risk

F(4, 27) = 2.90 p < 0.05 partial η2 = 0.18

op

Speed * Age

fC

effects for urban roads, including significant main effects of age for feeling of risk and task difficulty and a significant interaction of speed and age for the estimated probability of a crash (Table 3.5). Games Howell post-hoc tests were calculated for the two main effects of age found for feeling of risk and task difficulty. They indicated that older participants’ feelings of risk were significantly higher than those of young participants across all speed and risk conditions (meandiff = 1.24, p < 0.05). For task difficulty, the post-hoc test just failed to reach significance, but indicated that older participants’ ratings of task difficulty were somewhat higher than those of young participants across all speed and risk conditions (meandiff = 1.17, p = 0.053). A significant interaction between age and speed emerged for estimates of the probability of a crash. In the low speed condition, crash probability estimates were similarly low for all three age groups. However, whereas crash probability estimates of young and middle aged drivers were similar for the average and fast driving condition and showed moderate increases with speed, older drivers’ crash estimates for the average and fast speed condition, rose considerably more steeply.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

33

Preferred and maximum speeds Age-related differences between preferred and maximum speeds were calculated using the data from the free drive condition (second part of the trial). Findings from separate split plot ANOVAs (Table 3.6) for the two road environments indicated significant main effects for speed and age and a significant interaction for speed and age on urban roads. For the dual carriageway, significant main effects for speed and risk were found, as well as a significant interaction for risk and age.

9781472414694_Dorn.indb 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:00 AM


34

Table 3.6

Significant findings from split plot ANOVAs for age differences 1 2 in driven speed in the second part of the simulator study ANOVA results for adopted speeds

Urban road

F(1, 24) = 72.71, p < 0.001, partial η2 = 0.75

Age

F(2, 24) = 13.07, p < 0.001, partial η2 = 0.53

Speed * Age

F(2, 24) = 4.61, p < 0.05, partial η2 = 0.26

Speed

F(1, 24) = 62.77, p < 0.001, partial η2 = 0.72

Risk

F(1, 24) = 34.17, p < 0.001, partial η2 = 0.59

Risk * Age

F(2, 24) = 3.97, p < 0.05, partial η2 = 0.25

y

Dual carriageway

Speed

oo

fC

op

An ordinal interaction between speed and age on urban roads (Figure 3.3) indicated that preferred speeds were significantly lower than maximum speeds and that both speeds decreased with increasing age. Bonferroni post-hoc tests for the significant main effect of age indicated that young drivers chose significantly higher speeds than middle aged (meandiff = 9.77, p < 0.01) and older drivers (meandiff = 15.06, p < 0.001). For the dual carriageway a significant main effect for speed indicated that preferred speeds were always lower than maximum speeds. The interaction between risk and age, depicted in Figure 3.4, showed that adopted speeds were always lower in the presence of other oncoming road users, but that this difference

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Figure 3.3

9781472414694_Dorn.indb 34

Estimated marginal means plots for the significant interaction between speed and age on urban roads

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:01 AM

op

Estimated marginal means plots for the significant interaction between risk and age on urban roads

fC

Figure 3.4

was most pronounced for the oldest drivers, whose mean adopted speeds dropped below those of the middle-aged drivers when other road users were present and rose above them when there were none. Accuracy of speed estimates

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

35

y


Differences between actual speed and reported speed were fed into a split plot ANOVA with age as a between-subject factor and risk, and preferred versus maximum speed as within-subject factors. Significant age effects were found for both road environments (Table 3.7). Table 3.7

Significant findings from split plot ANOVAs for differences in the accuracy of speed perceptions in the second part of the simulator study ANOVA results for speed difference (actual – perceived)

Urban road

Age

Dual carriageway Speed

9781472414694_Dorn.indb 35

Risk

F(2, 24) = 14.12, p < 0.001, partial η2 = 0.54 F(1, 24) = 17.98, p < 0.001, partial η2 = 0.43 F(1, 24) = 33.86, p < 0.001, partial η2 = 0.59

Risk * Age F(2, 24) = 8.70, p < 0.001, partial η2 = 0.42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:01 AM


op

fC

Figure 3.5 Estimated marginal means plots for the main effect of age on urban roads

oo

On urban roads, Bonferoni post-hoc tests for the significant main effect of age indicated that young drivers estimated their driven speed differently from middle-aged (meandiff = 9.43, p < 0.001) and older drivers (meandiff = 8.16, p < 0.001); whilst young drivers slightly underestimated their driven speed on urban roads, middle-aged and older drivers considerable overestimated their speed in this environment (see Figure 3.5). Significant main effects for speed and risk on the dual carriageway were found. Preferred speed estimates were more accurate than maximum speeds in this road environment, and speed estimates were less accurate if no other road users were present (generally underestimations of speed). The significant interaction between age and risk indicated that speed assessments for all age groups were more accurate when other road users were present in the scene (see Figure 3.6). Older drivers’ estimates were least accurate and middle-aged drivers’ estimates were most accurate for both risk conditions.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

y

36

Discussion The present study tested predictions of Fuller’s Risk Allostatis Model and expanded on previous replication studies. Based on the model assumption that drivers continuously monitor the difficulty of the driving task (through monitoring feelings of risk) rather than the probability of a crash, Fuller posited a threshold

9781472414694_Dorn.indb 36

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:02 AM


y op

fC

Figure 3.6 Estimated marginal means plots for the main effect of age on the dual carriageway relationship for drivers’ subjective likelihood of a crash and linear increases for task difficulty and for feelings of risk with increasing speeds. Task difficulty and feeling of risk

Task difficulty and feeling of risk were found to be strongly correlated with each other. Whilst Fuller observed correlations of r = 0.98, Kinnear found them to range between r = 0.71–0.79. Lewis-Evans and Rothengatter reported correlation coefficients that ranged between r = 0.81–0.91. The present study confirmed the posited relationship between task difficulty and feeling of risk, with correlation coefficients ranging between r = 0.70–0.87. In contrast to Kinnear et al. (2008), who observed increases in the strength of the correlations with ascending speeds, this was only observed for the dual carriageway environment in the present study. For urban roads, the opposite was the case. As predicted by the model, correlations of feeling of risk and estimates of the likelihood of a crash were considerably weaker.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

37

Speed as the main determinant of task difficulty Previous studies have demonstrated repeatedly that, in line with model predictions, drivers’ ratings of task difficulty are closely associated with speed. In the current study speed also emerged as the most important predictor of task difficulty; however, the amount of variance explained was considerably lower than that

9781472414694_Dorn.indb 37

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:03 AM


38

y

reported by Fuller et al. (2009). The values were closer to, however somewhat lower than, those reported for residential roads and dual carriageways in LewisEvans and Rothengatter’s (2009) study. Furthermore, the inclusion of other road users in the current study impacted ratings of task difficulty (and feeling of risk) significantly, albeit weakly on the dual carriageway. Compared with the urban roads, where other road users were simulated as oncoming vehicles (and thus without a potential impact on the driver’s actions), other road users on the dual carriageway were simulated as vehicles in the adjacent lane (thus with the potential impact on the driver’s actions). We suggest that research to date has focussed exclusively and artificially on one determinant of task difficulty and that further research should explore other potential drivers of task difficulty.

op

Age effects

oo

fC

The current study asserted a specific interest in the exploration of age effects and age-related changes on the central variables of Fuller’s Risk Allostasis Model, including ratings of task difficulty, feeling of risk and likelihood of a crash. Furthermore, the study explored age effects for the accuracy of participants’ task difficulty perceptions by comparing drivers’ perceived speed to the actual speed driven and for adopted speed in a free drive condition. The findings for urban roads suggested that older drivers experienced stronger feelings of risk and greater task difficulty in the urban environment, and that their estimates of crash likelihood increased more steeply with ascending speeds than those of middle-aged and younger drivers. This was accompanied by older drivers’ preference for significantly lower preferred and maximum speeds on urban roads in the free drive condition, compared with the youngest age group. Similarly to middle-aged drivers, older drivers overestimated their speed in the free drive condition on urban roads, whereas young drivers underestimated speeds, but to a lesser degree than the other two age groups. No significant differences between drivers of different age groups were found for feeling of risk, task difficulty and estimates of crash likelihood on the dual carriageway. Older drivers’ speed assessments were always more inaccurate than those of middle aged and young drivers on the dual carriageway, but were relatively more accurate in drives where other road users were present. In this road environment mis-assessments of speed were underestimations of speed for all age groups with middle-aged drivers being most accurate. Somewhat surprisingly older drivers adopted speeds on dual carriageways that were higher than those of middle aged drivers, but lower than those of young drivers, if no other road users were present. In the presence of other road users, however, older drivers’ speeds were lower than those of middle-aged and young drivers. The combined findings from the first and second part of the study suggest that drivers broadly chose speeds to fit their perceived difficulty of the task. The fact that chosen speeds seem to map onto the findings for speed perception accuracy

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:03 AM


References

fC

op

y

suggests that the speeds chosen were affected by misperception of the speeds in the first place. It is noteworthy that whilst young drivers underestimated speeds in both environments, middle-aged and older drivers overestimated speed on urban roads and underestimated them on the dual carriageway. The finding of significantly higher feelings of risk, task difficulty and crash risk in the urban environment and the fact that older drivers modified their adopted speed and speed estimates in reaction to the presence of other road users on the dual carriageway could point to the use of different perceptual cues in this age group. The current study cannot provide a detailed breakdown of the factors on which drivers base their ratings of task difficulty and feeling of risk. We, however, suggest that further research is needed into the constituent components of perceived task difficulty and its relation to objective task demand. Only if perceived difficulty correctly reflects objective task demand can drivers make driving decisions that enable them to make progress whilst maintaining safe control over their vehicle. Kuiken and Twisk (2001, p.14) were the first to describe calibration as “the ability of a driver to recognise the relationship between the demands of the driving task and their own abilities, including error recovery. At any moment in time, a driver needs to be actively engaged in assessing what the driving task requires in terms of actions or the avoidance of actions, and the potential difficulties involved”. If it is accepted that correct calibration is an essential element of safe driving for young novice drivers, it can also be argued that it should also apply to older drivers and drivers of all ages.

Carsten, O. (2007). From Driver Models to Modelling the Driver: What do we really need to know about the driver. In P. Cacciabue (ed.), Modelling driver behaviour in automotive environments: Critical issues in driver interactions with intelligent transport systems (pp. 105–21). London: Springer-Verlag. Department for Transport. (2008). Road Statistics 2007: Traffic, Speeds and Congestion. London: Department for Transport. Fuller, R. (1984). A conceptualization of driving behaviour as threat avoidance. Ergonomics, 27(11), 1139–55. Fuller, R. (2000). The Task-Capability Interface model of the driving process. Recherche – Transports – Sécurité, 66, 47–59. Fuller, R. (2005a). Driving by the seat of your pants. A new agenda for research. Paper presented at the Behavioural Research in Road Safety: Fourteenth Seminar, London. Fuller, R. (2005b). Towards a general theory of driver behaviour. Accident Analysis and Prevention, 37, 461–72. Fuller, R. (2009). Recent developments in driver control theory: From task difficulty homeostasis to risk allostasis. Unpublished Manuscript. Dublin: School of Psychology, Trinity College Dublin.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

39

9781472414694_Dorn.indb 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:03 AM


40

oo

fC

op

y

Fuller, R., McHugh, C., and Pender, S. (2008). Task difficulty and risk in the determination of driver behaviour. Revue Européenne de Psychologie Appliquée/European Review of Applied Psychology, 58(1), 13–21. Haight, F. A. (1986). Risk, especially risk of traffic accident. Accident Analysis and Prevention, 18(5), 359–66. Johannson, G., and Rumar, K. (1966). Drivers and road signs: A preliminary investigation of the capacity of car drivers to get information from road signs. Ergonomics, 9, 57–62. Johansson, G., and Backlund, F. (1972). Drivers and road signs Applied Ergonomics, 3(1), 53. Kinnear, N., Stradling, S.G., and McVey, C.J. (2008). Do we really drive by the seat of our pants? In L. Dorn (ed.), Driver behaviour and training. Volume III. Aldershot, UK: Ashgate. Kinnear, N. (2009). Driving as you feel: A psychological investigation of the young driver problem. Unpublished doctoral thesis, Napier University, Edinburgh, Scotland. Kuiken, M., and Twisk, D. (2001). Safe driving and the training of calibration. Leidschendam: SWOV Institute for Road Safety Research. Lewis-Evans, B., and Rothengatter, T. (2009). Task difficulty, risk, effort and comfort in a simulated driving task—Implications for Risk Allostasis Theory. Accident Analysis and Prevention, 41(5), 1053–63. Näätänen, R., and Summala, H. (1974). A model for the role of motivational factors in drivers’ decision-making. Accident Analysis and Prevention, 6(3–4), 243–61. Näätänen, R., and Summala, H. (1976). Road User Behaviour and Traffic Accidents. Amsterdam and New York: North Holland/American Elsevier. Summala, H. (1988). Risk control is not risk adjustment: The zero-risk theory of driver behaviour and its implications. Ergonomics, 31(4), 491–506. Summala, H. (1996). Accident risk and driver behaviour. Safety Science, 22(1–3), 103–17. Summala, H. (2000). Automatization, automation, and modeling of driver’s behavior. Recherche – Transports – Sécurité, 66(3), 35–45. Summala, H., and Näätänen, R. (1988). The Zero Risk Theory and overtaking decisions. In J.A. Rothengatter and R.A. deBruin (eds), Road user behaviour (pp. 82–92). Assen/Maastricht: Van Gorcum. Taylor, D.H. (1964). Drivers’ galvanic skin response and the risk of accident. Ergonomics, 7(4), 439–51. Wilde, G.J.S. (1982). The Theory of Risk Homeostasis: Implications for Safety and Health. Risk Analysis, 2(4), 209–25. Wilde, G.J.S. (1988). Risk homeostasis theory applied to a fictitious instance of an individual driver’s decision making. In J.A. Rothengatter and R.A. DeBruin (eds), Road user behaviour (pp. 66–76). Assum/Maastricht: Van Gorcum.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 40

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:03 AM


41

Pr oo

fC

op

y

1 Wilde, G.J.S. (1989). Accident countermeasures and behavioural compensation: 2 The position of risk homeostasis theory. Journal of Occupational Accidents, 3 10(4), 267–92. 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 41

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:03 AM


5/30/2013 10:20:03 AM

Chapter 4

Development and Evaluation of a Competence-based Exam for Prospective Driving Instructors Erik Roelofs*, Maria Bolsinova*, Marieke van Onna* and Jan Vissers**

Introduction

y

Cito, National Institute for Educational Measurement, The Netherlands; ** Royal Haskoning DHV, The Netherlands

op

*

fC

A growing consensus among driver training and road safety researchers is that driver training should place greater emphasis on higher-order, cognitive and motivational functions underlying driving behaviour (Hatakka et al., 2002; Mayhew and Simpson, 2002). This changed conception of driver training has been laid down in the Goals for Driver Education matrix (Hatakka et al., 2002) and recent research seems to support this idea (Beanland, Goode, Salmon, and Lenné, 2013; Isler, Starkey, and Sheppard, 2011). Innovative training initiatives appear to counteract overconfidence and address motivational factors such as driving anger, sensation seeking, and boredom (e.g., Isler et al., 2009). Parallel to the doubts raised about the quality of driver training, the quality of driver instructor preparation programmes have been criticised. The MERIT review study (Bartl, Gregeresen, and Sanders, 2005) showed that huge variations existed in the quality of driver instructor education throughout Europe. The content often did not cover higher order skills and most programmes relied on teacher-focused approaches, which seem to fall short in developing higher order skills. In many European countries the quality of the education of driving instructors is regulated using an instructors’ exam. One may view this as a problem, but on the other side, this also offers opportunities for improvement. A valid and reliable exam that only allows proficient prospective instructors to enter the profession, may have a positive backwards effect on driver instructor education programmes, as in other fields of education: teachers teach and students learn what will be tested (Crooks, 1988; Fredericksen and Collins, 1989; Madaus, 1988). In the Netherlands, the first steps in this direction have been made in the last ten years. As part of a new law on driving education, in 2003 competence-based outcome standards for prospective driving instructors were formulated (Nägele, Vissers, and Roelofs, 2006). The most far-reaching change underlying these standards has been the emphasis on performance in critical job-situations with

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 43

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:03 AM


44

op

y

real learner drivers. In addition, supporting knowledge was defined in terms of relevant concepts, principles and decision making skills to be applied in authentic instructional situations. Based on the standards, a two stage competence-based exam was designed and put into action in the fall of 2009. Since then, over 4,000 prospective driving instructors (PDIs) have gone through one or more tests. The question is whether the assessments have resulted in valid and fair decisions about the suitability of PDIs. Regarding the tenability of decisions, this paper focuses on the separate theoretical assessments, comprising stage 1. In addition, their predictive value for instructor performance, as demonstrated during the final performance assessment lesson (stage 2), will be investigated. The fairness question concentrates on the comparability of different versions of the assessments. In the exam, item banks are used, from which different sets are drawn to compose exam versions to reduce practice effects and cheating. The question then arises whether one cut-off score implies the same level of required proficiency for the different versions. To solve this problem, psychometrical equating methods are commonly used to determine how scores on two different tests can be projected on to one (latent) scale (Kolen, and Brennan, 1995). In summary, four research questions are investigated here:

oo

fC

1. To what extent are the individual parts of the exam psychometrically reliable? 2. To what extent do the different theoretical tests interrelate? 3. To what extent do results on the theoretical tests and the performance assessment for instructor ability correlate? 4. Do the cut-off scores used across different versions of the theoretical tests reflect equivalent levels of proficiency? Design Features of the Competence-based Exam

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

The exam consists of two stages. The first stage consists of the assessment of the theoretical knowledge base of the prospective instructors regarding driving and driving pedagogy. After having passed the first stage PDIs receive a provisional instructor licence enabling them to enrol in a half year internship at a certified professional driving school. In the second stage, after having finished their internships, PDIs are judged on their professional instructional abilities, during a masterpiece lesson involving one of their own learner drivers, whom they have been teaching as an intern. If they pass, they will be granted a full licence for the next five years. The design features are described below and a summary of all exam sections is provided in Table 4.1.

9781472414694_Dorn.indb 44

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:03 AM

9781472414694_Dorn.indb 45

5/30/2013 10:20:03 AM

1.2 Verbalising mental processes of driving The driving instructor is able to verbalise the mental task processes that take place when carrying out driving tasks in different traffic situations.

2.2 Elaborating driving pedagogy The driving instructor is able to prepare a driving specific pedagogical learning environment for learner drivers.

2.1 Adaptive planning The driving instructor is able to construct an educational program for the long term (curriculum) and for the short term (lesson design) adapted to the needs of the individual learner driver (LD).

1. Competent in conscious traffic participation

2. Competent in lesson preparation

Elaboration of task domains

2.3 Organising learning The driving instructor is able to organise lessons in such a way that activities run smooth and without interruptions, ensuring a maximum amount of productive learning time.

continued ...

Performance assessment drive as a first and a second driver (all task domains cluster 1)

Theory of driving test (60 items): traffic participation rules knowledge, case-based and situational judgement items (task domain 1.1)

Theory of lesson preparation test (60 items): case-based concept application, reasoning and situational judgement items (all task domains cluster 2)

Stage 2: On the job: Performance assessment

Assessment method Stage 1 Computer based: Knowledge base and cognitive skills

Instructor competence profile and tests used for the Dutch driver training exam

1.1 Driving responsibly as a first driver The driving instructor is able to drive a vehicle safely, smoothly, be socially considerate, and in an eco-friendly way, according to Dutch driving standards.

Table 4.1

9781472414694_Dorn.indb 46

5/30/2013 10:20:03 AM

4.2 Reflection and revision The driving instructor is able to reflect on their own actions and use the results of this reflection for adapting their approach.

4.1 Assessing learner progress The driving instructor is able to assess the progress in driver competence by judging their level of performance and by using the expertise of professional colleagues.

3. Competent in instruction and coaching

4. Competent in evaluation, reflection and revision

Elaboration of task domains

3.2 Providing coaching The driving instructor is able to monitor learner driver development and guide the LD towards self-regulation in solving driving tasks and driving related tasks.

concluded

3.1 Providing instruction The driving instructor is able to provide instruction that is geared to the actual developmental level of the learner driver. It enables the LD to progress towards selfregulated performance in increasingly complex tasks.

Table 4.1

1. Performance assessment lesson with real learner driver 2. Self-reflection report internship 3. Reflective interview internship (all task domains cluster 2,3 and 4)

Theory of instruction and coaching test (60 items): case-based concept application, reasoning and situational judgement items (all task domains cluster 3)

Stage 2: On the job: Performance assessment

Stage 1 Computer based: Knowledge base and cognitive skills

Assessment method

Competence-based Exam for Prospective Driving Instructors

The design process

The conceptual assessment framework

y

All individual assessments of the exam were designed by means of the evidencecentred design model (ECD, Almond, Steinberg, and Mislevy, 2002; Mislevy, and Haertel, 2006). The ECD model identifies five layers in the design process: domain analysis, domain modelling, conceptual assessment framework, assessment implementation, and assessment delivery. These design layers were implemented successively, whereby a continuous dialogue took place between the assessment designers and different stake holders, which included: a board of instructor educators, exam institutes, ICT specialists, psychometricians, educational scholars, academics specialising in education and teaching, driving examiners, and driving instructors.

fC

op

The layer of the conceptual assessment framework for the assessment of task design was of central importance in the exam. The conceptual assessment framework helps to sort out the relationships among attributes of a candidate’s competence, observations which show competence and situations which elicit relevant driver performance. The central models for task design are the Competence or Student Model, the Evidence Model, and the Task Model. The driving instructor competence model

The Competence Model encompasses variables representing the aspects of instructor competence that are the targets of inference in the assessment and their inter-relationships. A competence model was constructed starting from a literature search on what comprises good teaching in general and more specifically driving instruction. This resulted in the formulation of four domains of competence, as summarised in Table 4.1: 1) Conscious traffic participation as first and second driver; 2) Lesson preparation; 3) Instruction and coaching; and 4) Evaluation, reflection and revision. A model of competent task performance has formed the basis for two competence models: driving competence (Roelofs, van Onna, and Vissers, 2010) and instructional competence (Roelofs and Sanders, 2007). A basic tenet in the model (see Figure 4.1) is that instructor competence is reflected in the consequences of an instructor’s actions. The most important consequences are students’ learning activities during the delivery of instruction or coaching, and safety, traffic flow and comfort in traffic whilst driving. Starting from the consequences, the remaining elements of the model can be mapped backwards. First, the component, actions, refers to professional activities being performed (e.g., delivering instruction or providing coaching to students). Second, any instructor activity takes place within a specific context in which many decisions need to be made, on a long-term basis (planning ahead)

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

47

9781472414694_Dorn.indb 47

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:03 AM


Model of competent task performance

op

Figure 4.1

fC

or immediately during an in-car situation. For instance, instructors will have to plan their instruction and adapt it depending on differing circumstances (e.g., different learning paces, traffic situations). Thirdly, when making decisions and performing activities, teachers have to draw from a professional knowledge base (e.g., pedagogical principles, psychology of driving, rules and regulations). Assessment task models and exam construction

oo

The Task Model describes the kinds of assessment tasks (items) that embody an assessment (test). They follow directly from the cognitive activity and interactive activity, as mentioned in the competence model. Three types of assessment tasks were designed: 1) Case based items (Schuwirth et al., 2000; Norman, Swanson and Case, 1996). These items address knowledge of concepts and cause-effect rules in cases embedded in a rich driving instruction context. These questions were used for theoretical assessment in all four task domains. An example of an item is “An instructor starts a lesson with an explanation of the first topic. The instructor does not outline what the learner driver is going to be able to do at the end of the lesson. What is the most likely consequence of this approach?” To respond, the PDI would choose from four options: A) the learner will learn less than is desirable, B) the learner will not fully understand your explanation, C) the learner will have less time to practice new driving tasks, D) the learner cannot direct their attention to the essential parts of the lesson (correct answer). 2) Situational judgement items (Whetzel and McDaniel, 2009). These items address decision making skills in a rich driving instruction context. An example regarding lesson planning is: “The learner driver is starting to

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

y

48

9781472414694_Dorn.indb 48

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:04 AM


op

y

learn manoeuvring in traffic situations with low traffic density, as illustrated on the picture. During the upcoming lesson you are going to instruct them in how to park backwards into a parking bay. Which of the parking bays [pictures shown with or without other vehicle parked alongside an empty bay] would be the best choice for a learner driver in this stage of driver education?” To respond, the PDI would choose one out of the four pictures. One option is optimal, one is suboptimal, and the remaining two are unacceptable. 3) Performance assessment assignments. To address the PDIs own driving competence and ability to verbalise mental task processes (perception, anticipation, decision, action and execution) the PDI had to complete a 60 minute drive, on which their performance was judged by a trained assessor using a standardised scoring form. Five performance criteria were provided on this form: 1) safety, 2) aiding traffic flow, 3) driving in a socially considerate way, 4) ECO friendly driving, and 5) vehicle control (Roelofs, Van Onna, and Vissers, 2010). At two intermediate stops the PDI was asked to verbalise the mental processes that went on while solving the previous traffic situations.

fC

Finally, professional actions regarding instruction, coaching and evaluation were judged by having the PDI carry out a lesson with a real learner driver. The lesson performance will be judged by trained assessors. To this end a 34-item scoring form will be used to judge the quality of coaching and instruction. The Evidence Model describes how to extract the key items of evidence from instructor behaviour, and the relationship of these observable variables to the competence model variables. All three theory tests consisted of 60 multiple choice items. The Theory of Driving Test consisted of one-best-answer type items, scored dichotomously (0.1), and the cut-off score for passing the test was 42. Both the Theory of Lesson Preparation Test and the Theory of Instruction and Coaching Test included situational judgement items, which were assessed using the partial credit model. The best answer yielded seven points, a suboptimal answer yielded three points, while the distractors yielded no points. The case-based knowledge items had one best answer; all items were scored zero and seven, for incorrect and correct answers respectively. The cut-off score for passing this test was 266 points (out of 420 points). The items on the Performance Assessment Lesson were scored on a three point scale, representing counterproductive performance, beginning productive performance and optimal performance. A detailed scoring guide was available for assessors. Initial rater agreement scores using Gower’s similarity index (Gower, 1971) showed acceptable levels of agreement (0.67 for instruction and 0.75 for coaching). The cut-off score for passing the performance assessment was 71 points (out of 102 points), under the condition that the results were not below the specified cut-off score (M = 2.0) for no more than one out of seven categories.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

49

9781472414694_Dorn.indb 49

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:04 AM


50

Method Participants

oo

Analyses

fC

op

y

All data from candidates who enrolled on the programme between 1 January 2010 and 1 October 2012 were selected. Data from 2009 were discarded because the computer-based assessment platform was not completely stable at that time. This resulted in assessment data for 4741 prospective driving instructors. The PDIs taking part in the present study were 79 per cent male and 21 per cent female. The mean age was 34.9 years (SD = 10.9). Of these most 3079 (74.4%) were born in the Netherlands with the remaining 25.6 per cent coming from 79 different countries. Of those not born in the Netherlands, the majority were immigrants from Morocco (n = 199), Suriname (n = 190), Turkey (n = 151), Afghanistan (n = 112), Iraq (n = 89), and Iran (n = 46). A total of 4644 PDIs completed at least one of the theory tests. Of these 2977 passed all their theory exams. Of these, 1941 PDIs took part in the Performance Assessment Lesson assessing instruction and coaching. From the remaining PDIs about half (n = 508) did not participate in the Performance Assessment Lesson within a year of their last successful theory test. The remaining participants (n = 528) did not finish their internship. A total of 368 PDs received dispensation to participate in the performance assessment, although they failed one of the theory tests. In total, 2,315 PDIs participated in the Performance Assessment Lesson at least once.

Item response theory (IRT) analyses were performed on the data of the three theory tests. Each test had been administered in many different versions, based on the 60 item samples drawn from an item bank. The available software does not allow IRT-analysis on a very large number of versions of a test (there were more than 700 versions for each test), especially if some of these versions were completed by only one PDI. Therefore, for each of the theory tests only versions with a substantial number of participants were selected for analyses. In Table 4.2 the number of test versions, total number of items and sample size chosen for analysis are shown. For the tests which included items with partial scoring (0, 3, 7), a partial credit model was applied to the data first. The model had a very poor fit in both cases. Responses to all items in the tests Theory of Instruction and Coaching and Theory of Lesson Preparation were dichotomised, because this allows more flexibility to fit the data to the IRT model better. All items for which seven points were awarded were recoded as one (correct response), and all others as zero (incorrect response). The number of items correct had a very high correlation with the original number of points (0.986 for both tests). A new cut-off score for the number of items correct was chosen in such a way that the number of participants

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 50

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:04 AM


Table 4.2

Number of test versions, total number of items and sample size chosen

Test

Minimum Maximum Total Number of Sample number of number of number of test versions size responses per responses per items selected version version 14

99

484

211

3013

Theory of Lesson Preparation

15

38

586

201

2524

Theory of Instruction and Coaching

15

32

551

148

2771

y

Theory of Driving

fC

op

having less than 266 points, but passing the test, and the number of participants having more than 266 points, but failing the test, according to the new criterion was minimal. For both tests, the new cut-off score of 35 items correct resulted in 4.65 per cent misclassifications for the test of Theory of Instruction and Coaching and 3.95 per cent for the test of Theory of Lesson Preparation. A one parameter logistic model (OPLM, Verhelst, and Glas, 1995; Verhelst, Glas, and Verstralen, 1995) was fitted to the data for each theory test. In the Theory of Driving Test, one of the items was excluded from the analysis because it was answered correctly by all PDIs. In the Theory of Instruction and Coaching Test, two items were excluded because they had strong negative correlations with the test score. The model had a reasonable fit for all three tests. In terms of the IRT, the concept of reliability differs from that of the Classical Test Theory. The measurement precision depends on the position on the latent ability scale. The ability of a participant for whom all test items were too difficult was measured with a larger error than the ability of someone for whom the difficulty of the items was appropriate. In the case of the three theory tests, it is important to accurately measure the ability level corresponding to the cut-off score, represented by the pass-fail boundary. A standard error of estimate of ability, corresponding to the required number of items correct (SE), was chosen as a measure of reliability. To enable interpretation of this value it has to be compared with the standard deviation of the ability in the population (SD). As a measure, we look at the so called proportion of true variance: (SD2 – SE2)/SD2. Additionally, as a measure of global accuracy of measurement the MAcc index was used, reported using OPLM software, which represents the expected reliability coefficient alpha, as defined in classical test theory for a given population of PDIs. Using an IRT model enables all different versions of the test to be placed on the same scale and therefore different versions can be compared according to their difficulty and the level of ability needed to pass the test. For each test version a level of ability was estimated at the cut-off score. Unlike the raw test scores, the

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

51

9781472414694_Dorn.indb 51

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:04 AM


52

op

y

latent estimates from different test versions are directly comparable. All latent abilities were scaled in such a way that the population distribution had a mean of 100 and standard deviation of 15. To analyse relations between the three theory tests, a subsample of participants (n = 1980) who had taken all three tests were selected. Correlations between scores for the three latent abilities were computed. To assess the reliability of the final Performance Assessment Lesson, all available raw data (n = 580) for PDIs were obtained and processed into data files. The exam institute only kept the final pass/fail outcome in their data files, but still had score forms for 580 PDIs. Principal Component Analysis was carried out to reduce the data to a limited number of interpretable factors, yielding maximum reliable criterion variables. Correlations between three latent abilities and the score on the resulting scales were computed for instruction and coaching. Results

oo

fC

Figure 4.2 shows that the ability scores were normally distributed (M = 100; SD = 15). In this figure the cut-off scores for the 14 most frequently administered versions of the Theory of Driving Test are plotted. The dots represent the cut-off scores for each test version, expressed in terms of ability (Theta) and the lines represent the standard errors around the cut-off score. Two points can be noted here. First, the cut-off score for all test versions fall below the average ability in the total population. The mean cut-off score (M = 85.3, see Table 4.3) is almost one standard deviation below the ability mean of the population. This means that relatively low ability (M = 86.5) was needed to pass the test. Second, there are Table 4.3

Cut-off scores and cut-off reliability for the three theoretical tests

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Cut-off Score

Reliability Average proportion of true variance at the cut-off score

MAcc

SD

Min

Max

Average SE at the cutoff score

Theory of Driving

86.5 3.22

81.9

92.9

10.01

0.55

0.70

Theory of Lesson Preparation

85.3 5.47

77.3

93.1

9.56

0.59

0.75

Theory of Instruction and Coaching

90.2 2.52 86.93 95.5

7.74

0.73

0.83

M

9781472414694_Dorn.indb 52

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:04 AM

op fC

Figure 4.2

Cut-off scores and standard errors for 13 versions of the Theory of Driving Test

small differences between the required ability levels for different test versions, but the variation of the cut-off levels (SD = 3.22) across versions is small compared to the standard error. Similarly, Figure 4.3 (below) shows the plotted ability cut-off scores for the 15 most frequently administered versions of the Theory of Lesson Preparation Test. As can be noted from the figure the differences between the cut-off scores of the different versions of the Theory of Lesson Preparation Test were higher than for the Theory of Driving Test. Versions 1 and 6 differ considerably: Version 1 required an ability level of 78, while Version 6 required an ability level of 92. The mean cut-off score (M = 85.3) for Version 1 was again below the ability mean of the population. The results regarding the Theory of Instruction and Coaching Test show a similar pattern (see Figure 4.4 below). The required ability levels were again below the mean ability (M = 90.2, see Table 4.3), but this difference was less pronounced compared to the other theory tests. The different test versions also showed different levels of required ability, but these are relatively low compared to the standard error. For the Theory of Driving Test the standard error at the cut-off scores amounts to 10.01, whereas the values for the Theory of Lesson Preparation and Theory of Instruction and Coaching amount to 9.56 and 7.74, respectively (see Table 4.3).

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

53

y


9781472414694_Dorn.indb 53

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:07 AM


op fC

Cut-off scores and standard errors for 15 versions of the Theory of Lesson Preparation Test

oo

Figure 4.3

All values fall within the region of good reliability. The MAcc-coefficients reflect reasonable overall reliability for the whole test, whereas the average proportion of true variance at the cut-off score measures reflect questionable reliability for the first two tests and reasonable reliability for the Theory of Instruction and Coaching Test. The correlation coefficients between the ability scores on the three theory tests show a moderate correlation of 0.56 between the Theory of Driving Test and the Theory of Lesson Preparation Test. Ability scores on the Theory of Driving Test correlate 0.43 with the ability scores on the Theory of Lesson Preparation Test. Finally, ability scores on the Theory of Lesson Preparation Test correlate 0.29 with ability levels on the Theory of Instruction and Coaching Test. Table 4.4 shows the psychometric report for the Final Performance Assessment Lesson. Principal component analyses resulted in three clearly interpretable factors. Three fairly reliable scale scores were produced (see Table 4.4), representing aspects of Coaching and Motivational Support (alpha = 0.77), Diagnosis and Task Support (alpha = 0.79) and Instruction (alpha = 0.83). The Motivational Support Scale correlated 0.46 ( p < 0.001) with Diagnosis and Task Support and 0.45 ( p < 0.001) with Instructional skill. Diagnosis and Task Support correlated 0.66 ( p < 0.001) with Instructional Skill.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

y

54

9781472414694_Dorn.indb 54

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:10 AM


y op fC

Figure 4.4

Cut-off scores and standard errors for 15 versions of the Theory of Instruction and Coaching Test

Table 4.4

Psychometric report for the Final Performance Assessment Lesson

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

55

n

Minimum Maximum

M

SD

Alpha

Coaching: Motivational support (6 items)

580

9.00

18.0

14.3

2.2

0.77

Coaching: Diagnosis and task support (8 items)

580

9.00

24.0

16.9

2.8

0.79

Instructional skill (15 items)

580

21.00

44.0

34.7

4.3

0.77

Exam score (34 items)

580

50.00

99.0

78.2

8.9

0.88

Table 4.5 shows the correlations between the ability scores on the theory tests and the scores on the Performance Assessment Lesson, for the three subscales and the overall assessment score. The correlation coefficients for the Theory of Driving Test did not differ significantly from zero. The ability scores for Lesson

9781472414694_Dorn.indb 55

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:12 AM


56

Preparation and Instruction/Coaching show six low but significant correlations 1 with the subscales and the overall scale for the final performance assessment 2 3 lesson (between 0.12 and 0.14; p < 0.05). Table 4.5 Correlations between performance on theory tests and Performance Assessment Lesson Theory of driving

Theory of lesson preparation

Theory of instruction coaching

Coaching: Motivational support (6 items)

0.01

0.06

0.13*

Coaching: Diagnosis and support of task process (8 items)

0.07

0.12*

0.09

0.10

0.12*

0.11*

0.07

0.12*

0.14**

Exam score (34 items)

Discussion

fC

* p < 0.05, ** p < 0.01

op

Instructional skill (15 items)

y

Sub-scale Final Performance Assessment Lesson

oo

The central question in this study was whether decisions made about prospective driving instructors, as they follow on from their results on the theory tests, are valid and fair, as part of a new competence-based standards in the Netherlands. Firstly it can be concluded that the overall reliability of the estimated ability scores on the theory tests showed acceptable levels. The reliability around the cut-off scores were also acceptable, which seems most important, because it is here where the pass/fail decisions are made. The Final Performance Assessment (lesson) also showed acceptable reliability in terms of the alpha value. Second, the IRT models showed an acceptable fit, suggesting that the tests represent separable one-dimensional abilities. The moderate correlation between the Theory of Driving Test and the Theory of Lesson Preparation Test shows that knowledge of the traffic task is an important predictor of the knowledge and decision making regarding lesson preparation. To a lesser extent high (or poor) achievement on lesson preparation was related to similar performance on instruction and coaching. Third, the predictive value of theory test performance for in-car instructional and coaching performance was very low. The ability scores for lesson planning and instruction and coaching produced very low, although significant, correlations with in car-performance for coaching and instruction. An explanation for this finding may be that only those who passed the stage one theory exams were

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 56

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:12 AM


fC

op

y

allowed to go through the final assessment. In addition, the effect of half a year of internship may have washed out initial differences between PDIs. Regarding fairness, the question was whether the different versions of the theory tests required the same level of proficiency to pass. A first finding was that the cut-off scores for the pass-fail decisions for the theory tests were well below the average ability level of the population, implicating relatively low ability requirements. The Theory of Coaching and Instruction Test and the Theory of Driving Test had comparable cut-off scores across versions and were hence equivalent in their ability requirements. For Lesson Preparation there were larger differences in required ability across test versions, although the differences fell within the range of the standards errors. As far as the construction and delivery processes were concerned, some aspects need further attention. Many of these are related to the way the assessments were delivered. The exam is computer administered at an exam office, involving items banks from which different versions are drawn. To achieve representativeness, all versions need to reflect all sub domains (at least nine for each test), mental activities (perception, decision making, action, cause-effect reasoning, concept recognition), and critical situations (learner characteristics, stage of acquisition, traffic situation) are distinguished. The relatively small size of the item bank resulted in the frequent re-using of items, which may have led to overexposure of the items, and resulted in the lowering of item difficulty. In addition, we observed that some of the items had poor quality (i.e., low or highly negative item-test correlation coefficients), and extreme p-values (near zero or one). In the current examination practice, poor items were not excluded ad posteriori from the tests, because shorter test versions would not have been accepted by stakeholders. However, it would have been defendable to estimate ability levels based on a smaller cleaned subset of items which yield a more reliable but still representative score. An optimal approach to warrant acceptable item quality is to pre-test all items on a representative sample of target candidates before putting them into the item banks. This however seems problematic because of the risk of early item exposure. In addition, exam costs would rise. However, in general exam data can be used to redesign and improve the exam over time. Following the evidence-based design model of Mislevy and colleagues, many questions can be answered using this approach: does the competence model reflected in the IRT model fit? Are any changes needed? Do the cut-off scores represent what we want PDIs to know and to be able to do? Can certain item characteristics be traced back to the way the item was designed? In short, using an evidence-based design model, in combination with improvements made along the way, our decisions about prospective driving instructors can be improved. In future research we intend to take a closer look at other parts of the exam, the functioning of different item types, the way items are presented, the stimuli used

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

57

9781472414694_Dorn.indb 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:12 AM


58

in items, the responses that are asked and the way these are related to estimates of PDIs’ abilities. To evaluate the long term effects of the exam for instructional practice, learner driver improvements and crash involvement, longitudinal research will be necessary. In such a study one should take into account the quality of all subsequent educational interventions and related driver activities to determine whether there is a case for driver training (Beanland et al., 2013). References

oo

fC

op

y

Almond, R.G., Steinberg, L.S., and Mislevy, R.J. (2002). Enhancing the design and delivery of assessment systems: A four-process architecture. Journal of Technology, Learning, and Assessment, 1(5). http://www.bc.edu/research/ intasc/jtla/journal/v1n5.shtml. Bartl, G., Gregersen, N.P., and Sanders, N. (2005). EU MERIT Project: Minimum Requirements for Driving Instructor Training. Vienna: Institut Gute Fahrt. Bartl, G., Gregersen, N.P., and Sanders, N. (2005). EU MERIT Project: Minimum Requirements for Driving Instructor Training. Final Report. Vienna: Institut Gute Fahrt. Beanland, V., Goode, N., Salmon, P.M., and Lenné, M.G. (2013). Is there a case for driver training? A review of the efficacy of pre- and post-licence driver training. Safety Science, 51, 127–37. Crooks, T.J. (1988). The impact of classroom evaluation practices on students. Review of Educational Research, 58, 438–81. Fredericksen, J.R., and Collins, A. (1989). A systems approach to educational testing. Educational Researcher, 18(9), 27–32. Gower, J.C. (1971). A General Coefficient of Similarity and some of its Properties, Biometrics, 27, 857–71. Hatakka, M., Keskinen, E., Gregersen, N.P., Glad, A., and Hernetkoski, K. (2002). From control of the vehicle to personal self-control; Broadening the perspectives of driver education. Transportation Research Part F: Psychology and Behaviour, 5(3), 201–15. Isler, R.B., Starkey, N.J., and Sheppard, P. (2011). Effects of higher-order driving skill training on young, inexperienced drivers’ on-road performance. Accident Analysis and Prevention, 43, 1818–27. Isler, R.B., Starkey, N.J., and Williamson, A.R. (2009). Video-based road commentary training improves hazard perception of young drivers in a dual task. Accident Analysis and Prevention, 41, 445–52. Kolen, M.J., and Brennan, R.L. (1995). Test Equating. New York: Springer. Madaus, G.F. (1988). The influences of testing on the curriculum. In L.N. Tarner (ed.), Critical issues in curriculum (pp. 83–121). Eighty-Seventh Yearbook of the National Society for the Study of Education, Part I. Chicago: University of Chicago Press.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 58

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:12 AM


fC

op

y

Mayhew, D.R., and Simpson, H.M. (2002). The safety value of driver education and training. Injury Prevention, 8, 3–8. Mislevy, R., and Haertel, G. (2006). Implications of Evidence-Centered Design for Educational Testing. Menlo Park, CA: SRI International. Mislevy, R.J., Steinberg, L.S., and Almond, R.G. (2003). On the structure of educational assessments. Measurement: Interdisciplinary Research and Perspectives, 1, 3–67. Nägele, R., Vissers, J., and Roelofs, E.C. (2006). Herziening WRM: een model voor competentiegericht examineren. Eindrapport [Revision Law on motor vehicle education: a model for a competence based exam]. Dossier: X4691.01.001; Registratienummer: MV-SE20060477. Amersfoort/Arnhem: DHV, Cito. Norman, G., Swanson, D.B., and Case, S.M. (1996). Conceptual and methodological issues in studies comparing assessment formats. Teaching and Learning in Medicine, 8(4), 208–16. Roelofs, E., and Vissers, J. (2008). Toetsspecificaties theoretische deeltoetsen WRM-examen. [Specifications of the theoretical exams for driver instructor candidates] Arnhem: DHV. Roelofs, E.C., van Onna, M., and Vissers, J. (2010). Development of the driver performance assessment: informing learner drivers of their driving progress. In L. Dorn (ed.), Driver behaviour and training. Volume IV (pp. 37–50). Farnham, UK: Ashgate. Roelofs, E., and Sanders, P. (2007). Towards a framework for assessing teacher competence. European Journal for Vocational Training, 40(1), 123–39. Schuwirth, L.W.T., Verheggen, M.M., van der Vleuten, C.P.M., Boshuizen, H.P.A., and Dinant. G.J. (2000). Validation of short case-based testing using a cognitive psychological methodology. Medical Education, 35, 348–56. Verhelst, N.D., and Glas, C.A.W. (1995). The one parameter logistic model. In G.H. Fischer and I.W. Molenaar (eds), Rasch models: Foundations, recent developments and applications (pp. 215–39). New York: Springer. Verhelst, N.D., Glas, C.A.W., and Verstralen, H.H.F.M. (1995). OPLM: One Parameter Logistic Model. Computer program and manual. Arnhem: Cito. Wang, M., Haertel, G., and Walberg, H. (1993). Toward a knowledge base for school learning. Review of Educational Research, 63, 249–94. Whetzel, D.L, and McDaniel, M.A. (2009). Situational judgement tests: An overview of current research. Human Resource Management Review, 19, 188– 202.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

59

9781472414694_Dorn.indb 59

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:12 AM


5/30/2013 10:20:12 AM

fC

op

y

Part 2 Driver Behaviour and Driver Training

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 61

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:12 AM


5/30/2013 10:20:12 AM

Chapter 5

Identifying the Characteristics of Risky Driving Behaviour Christian Gold, Thomas Müller and Klaus Bengler Munich Technical University, Germany

y

Introduction

fC

op

Inappropriate driving behaviour, aggression and distraction are types of risky driving behaviour that increase the risk of becoming involved in a motor vehicle crash, which in turn may have dramatic consequences for those involved. By definition, “risky driving refers to those patterns of driving behaviour that place drivers at risk for morbidity and mortality and that involve legal violations” (Jessor, Turbin and Costa, 1997, p. 4). Therefore, risky driving refers not only to reckless and aggressive driving, but also to inattentive, distracted driving and driving with excessive fatigue. Inattention, illness, or sleepiness, for example, are some of the most important causes of accidents (Minoiu, Netto, Mammar and Lusetti, 2009), and the negative effects of distraction on the driving task need not be elaborated here. Previous research has found there are many different manifestations of risky driving. The most frequently mentioned are speeding and tailgating (e.g., James and Nahl, 2000; Sarkar, Martineau, Emami, Khatib and Wallace, 2000; Tasca, 2000; Jessor et al., 1997). Other examples include such things as: cutting in front of another car, weaving in and out of traffic or running red lights (e.g., Shinar, 1998; Sarkar et al., 2000; Lajunen and Parker, 2001; Tasca, 2000). There are several approaches for preventing the negative consequences of such risky driving. Firstly, enforcement of the traffic laws can be increased, in order to stop risky driving in the first place. This is a wellestablished procedure for decreasing risky driving, although it will not stop risky driving completely. Another method for reducing the number of accidents caused by risky driving is by identifying risky drivers in the immediate vicinity and adapting the driver’s behaviour to account for this. Knowledge about the driving style of other road users can provide useful information for advanced driver assistant systems. In order to warn other road users and thus prevent the occurrence of dangerous situations at an early stage, risky driving behaviour must be reliably identified from outside the vehicle. Data recorded in-vehicle by a risky drivers’ car is not currently available for surrounding vehicles, at least until car to car communication becomes common.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:13 AM


64

Therefore, currently the identification of a risky driving style, on the part of other road users, must be based upon extrinsic data only (i.e., the vehicles movements). In order to identify the different types of risky driving behaviour, first of all the hazards which constitute a risk must be identified (Hoyos, 1988). These hazards can be present as fixed, stationary, or mobile objects in the driver’s vicinity (Brown and Groeger, 1988). According to Brown and Groeger (1988), the two main inputs of risk perception are:

oo

fC

op

y

1 2 3 4 5 6 7 8 9 1. Information on the types of hazards in the traffic environment 2. Information on the joint abilities of the driver and vehicle to prevent that 10 11 hazardous potential from being transformed into an actual accident 12 Only if other drivers have knowledge of the potential hazard, in this case 13 represented by a risky driver, can they take steps to avoid it and to prevent that 14 potential hazard from being transformed into an actual accident. As risk can be 15 defined as “the ratio between some measure of adverse consequences of events 16 and some measure of exposure to conditions under which those consequences 17 are possible” (Brown and Groeger, 1988, p. 586), risk will be decreased when 18 exposure to the risky driver decreases. Furthermore, reducing other drivers’ 19 exposure to the potential hazard (the risky driver) also decreases the potential 20 21 hazards faced by the risky driver. Therefore, it is important to be able to identify the characteristics of risky 22 driving behaviour. Besides the very well-known and easily observable risk factors, 23 such as speeding or tailgating the vehicle in front, many other parameters must also 24 be taken into account when assessing risky driving behaviour. If the car detects 25 such forms of risky driving and reacts in an appropriate way, the driver may not 26 agree, because their judgement of the situation differs from that of the cars. The 27 vehicles’ assessment is based on facts from sources like accident statistics and 28 derived from partially marginal forms of dynamic measures (e.g., the standard 29 deviation of lateral position), whereas the rating of the driver includes additional 30 subjective factors which are not considered by the system. The present study was 31 conducted at the Institute of Ergonomics at the Technische Universität München to 32 fill this gap in the literature by identifying the subjective and objective parameters 33 34 of risky driving behaviour. 35 36 37 Method 38 A literature review was conducted in order to create a list of the many different 39 manifestations of risky driving. Based on this review several different parameters 40 were selected and programmed into a driving simulator. Following this process, 41 two driving simulator studies were conducted. The first study was designed 42 to investigate whether there were other variables which influence a drivers’ 43 perception of risk, which are not related to risky driving, such as the colour of 44

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 64

5/30/2013 10:20:13 AM

Identifying the Characteristics of Risky Driving Behaviour

the vehicle. In the second study, 30 participants were confronted with several different vehicles driving in a risky manner and were asked to rate the subjective risk associated with each situation. By varying the characteristics of the different risky drivers, participants were exposed to the different manifestations of risky driving behaviour. Differences between these types of risky driving behaviour were investigated using the participants’ self-reported perceptions of risk in each situation and compared with the participants’ self-reported driving style. Parameter selection

fC

op

y

The literature on the topics of aggressive or reckless driving, road rage, distraction, fatigue and also crash statistics from Germany and the USA were reviewed. The outcome was a list of 63 different risky driving behaviours. They were ranked according to their relevance for the present study and their impact on road safety. Due to the practicalities of conducting a driving simulator study, only a limited number of situations could be implemented. In other words, the more situations the participants had to rate, the longer the experiment would need to be and the worse their concentration and motivation would become. In addition, another important consideration in the selection process was whether the risky behaviours could be displayed in a recognisable manner in the driving simulator. This judgement was based upon whether the risky driving behaviour could be observed and easily identified from the participants’ point of view. Furthermore, as objects in the mirrors were less likely to be noticed, all risky driving was presented in front of the participant during the simulated drive. In addition, not every characteristic of risky driving could be reproduced in the driving simulator. For example, horn honking or headlight flashing could not be reproduced in the simulator. Table 5.1 (below) presents the seven risky driving behaviours which were included in this study.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

65

The driving simulator

The present study is based on a driving simulator experiment held in a static driving simulator. The simulator consists of a full vehicle mock-up and six projectors, creating a 180 degree view and allowing the use of the driving mirror as well as the two wing mirrors. The simulator is equipped with an intercom system, so the experimenter can verbally communicate with the participant while supervising in the control station. The benefit of presenting the risky driver in a driving simulator is that, while the participant is directly affected, there is no real danger as there would be in a real traffic situation. However, by using a video representation instead, the participant does not have to be concerned and the ratings may be lower than they would be in reality. As enhancing the simulation experience by representing acceleration would not be expected to provide better results, a static simulator was chosen.

9781472414694_Dorn.indb 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:13 AM


66

Table 5.1 Risky driving manoeuvres included in the simulated driving Definition

Source (examples)

Overtake

To overtake somebody in an inappropriate way

Tasca (2000) Lajunen & Parker (2001)

Tailgate

To drive without sufficient distance to the lead vehicle

James & Nahl (2000) Tasca (2000) Sarkar et al. (2000)

Swerving

To oscillate around the ideal track, often caused by inattention

Knappe et al. (2007) Pizza et al. (2004)

Lane change

To change lanes frequently

James & Nahl (2000) Shinar (1998) Sarkar (2000)

Low speed

To drive at an inappropriately low Taubman et al. (2004) speed Lajunen & Parker (2001)

Speeding

To drive too fast for the prevailing conditions or faster than allowed by law

Tasca (2000) James & Nahl (2000) Begg & Langley (2001)

Speed change

To change speed regardless of the traffic situation

James & Nahl (2000)

op

fC

Questionnaire

y

Parameter

oo

A total of three sections were used in this questionnaire. The first section collected information about the demographics, such as age, gender and driving experience. The second section of the questionnaire contained the Driving Practices Questionnaire (DPQ), which was developed to measure self-reported behaviour exhibited while driving a motor vehicle (Kidd and Huddleston, 1994). It serves to cluster drivers according to their self-reported driving style. Participants respond to ten statements on a five point Likert scale (1 = Never to 5 = Always) (see Figure 5.3). The higher the score on the DPQ the less safe a driver behaves in traffic situations. Previous research has found that, compared to individuals with a low risk score (20), those with a high score (40) were three times as likely to have prior traffic violations (Kidd and Huddleston, 1994). A difference in risk perception between safe and unsafe drivers was expected. Due to the small number of participants in the present study, two groups were used instead of the three used in the original research (Kidd and Huddleston, 1994). All participants with a DPQ score lower than the mean were allocated to the low risk group and those with a higher than average score were allocated to the high risk group. The final section of the questionnaire was a situation-based scale which consisted of questions that were asked while the participants were driving in the simulator. After each situation, participants were asked the following questions:

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 66

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:13 AM


1. How risky do you think the behaviour of the other road user was in the situation you just witnessed (1 = Not at all risky to 20 = Extremely risky)? 2. Why did you choose this score? Is there any specific manoeuvre, behaviour pattern, or other feature which particularly attracted your attention? The first question generates the data, while the second question ensures that the participant rates the right driving behaviour. A wide scale of 20 points was chosen to reduce the chances of the participants remembering the score they gave in similar situations. The drawbacks of using a wide scale, such as possible mental overload or that the whole scale may not be used, were accepted.

y

Pilot study

fC

op

The pilot study was conducted to examine whether vehicle colour, vehicle type, lane markings and roadside vegetation influence risk assessment. Furthermore, the pilot study served as a validation of the questionnaires to assess ambiguous items and other problems before undertaking the main study. For the pilot study, the four risky driving behaviours: swerve, tailgate, speed change, and overtake were selected. Each independent variable was varied along with one of the four risky driving behaviours, while all other factors remained the same. This resulted in 10 different situations being considered (see Table 5.2). Table 5.2 Situations pre-study Parameter

Variable

Characteristic 1 Characteristic 2

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

67

Characteristic 3

Tailgate

Vehicle colour

Red

Grey

Black

Swerve

Vehicle type

Compact car

Truck

Roadster

Speed change

Lane markings Lane markings

No lane markings

Overtake

Vegetation

Grassland

Wooded

Twenty participants took part in the pilot study. All ten situations were presented to each participant while they were driving in the simulator. A within-subject design was used and the order of the situations changed between the participants. Once the participants had experienced each situation they were then interviewed by the experimenter over the intercom system and prompted to answer the situationbased questionnaire.

9781472414694_Dorn.indb 67

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:13 AM


op

Pilot study results

fC

Figure 5.1 Results of the pilot study

oo

The mean age of the participants was 26.1 years old (SD = 9.05). The youngest was 18 years old and the oldest was 59 years. The results of the pilot study can be seen in Figure 5.1. Surprisingly, none of the comparisons showed any significant differences when tested using paired t-tests. The largest difference was between the truck and the roadster feature ( p = 0.143, r = 0.331). Trucks were therefore excluded from the main study.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

y

68

Main study

In the main study, the seven situations from Table 5.1 were developed into scenarios so that different manifestations of the same behaviours could be examined to quantify each behaviour. The resulting 14 situations (see Table 5.3) were split into nine rural road and five motorway situations. All situations consist of an approach section, a test section and an interview section. During the interview section the participants are prompted to answer the situation-based questions via the intercom system while they continued driving in the simulator. Whilst the situations were presented in random order, two manifestations of the same risky driving behaviour were not allowed to follow each other and the five motorway situations were presented together.

9781472414694_Dorn.indb 68

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:13 AM


Table 5.3

Implemented situations

Situation

Road

Description

Speed low

rural

The Risky Driver (RD) drives 30 km/h below speed limit

Speed change 70–110

rural

The RD varies speed between 70 and 110 km/h

Speed change 80–105

rural

The RD varies speed between 70 and 110 km/h

Speeding low

motorway

The RD drives 148 km/h, speed limit 130 km/h

motorway

The RD drives 184 km/h, speed limit 130 km/h

motorway

The RD changes lane three times to proceed faster

Swerve medium

rural

The RD oscillates three times across the lane markings

Swerve obvious

rural

The RD oscillates three times across the lane markings

Tailgate medium

rural

The RD drives 15 m behind a truck (speed 70 km/h)

Tailgate obvious

rural

The RD drives 3 m behind a truck (speed 70 km/h)

Tailgate aggressive

motorway

The RD drives 10 m behind another car on the left lane of the motorway (speed 152 km/h, speed limit 130 km/h)

motorway

The RD shortens the distance to the leading vehicle up to 5 m before reacting (speed 80 km/h)

rural

The RD overtakes the participant in front of a bend with poor visibility

Overtake bend

op

fC

Tailgate inattentive

y

Speeding obvious Lane change

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

69

Overtake cut

rural

The RD overtakes and cuts in front of the participant (distance 7 m)

Experienced drivers are significantly better than novice drivers at anticipating the potentially hazardous outcomes of a driving situation (Jackson, Chapman and Crundall, 2009). This is in agreement with Benda and Hoyos (1983) who also found driving experience had a strong influence on estimating hazards in traffic situations. Prior accident involvement has also been shown to strongly influence a driver’s perception of risk (Jackson et al., 2009). For these reasons, participants were required to have held a drivers’ licence for at least three years prior to taking part in the main study. Furthermore, no participants from the pilot study were allowed to participate in the main study.

9781472414694_Dorn.indb 69

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:13 AM


70

Results Participants Thirty subjects, who were mainly students or scientific assistants, participated in the main study (10% female). The mean age was 26.8 years (SD = 3.51) with a range of 21 to 33 years old. The majority of the sample (40%) reported an annual mileage of between 5,000 and 10,000 km per year. Twenty per cent stated that they drove less than 5,000 km per year. The other participants reported an annual mileage of between 10,000 to 20,000 km per year (23%) and more than 20,000 (17%).

y

Risk assessment

oo

DPQ score

fC

op

The shaded bars in Figure 5.2 show the mean risk scores for the different situations. The situations assessed as most risky were: overtake on a bend (M = 17.1; SD = 4.1), followed by aggressively tailgate (M = 14.2; SD = 4.19) and swerve obvious (M = 12.3; SD = 4.78) as well as tailgate obvious (M = 12.2; SD = 3.99). All the situations which were concerned with speed choice received relatively low ratings and were therefore only considered slightly risky. For example, speed change 70–110 km/h (M = 5.16; SD = 3.62), driving too slowly speed low (M = 4.23; SD = 2.85), and driving at 184 km/h speeding obvious (M = 3.37; SD = 3.37). A cluster analysis (Ward linkage) categorised the situations into three classes. Cluster 1 was characterised by a high risk rating, Cluster 2 with a medium risk rating and Cluster 3 with a low risk score (which included all of the speed situations).

The mean DPQ score was 27.5, which was similar to the findings for the DPQ score (M = 25) reported by Kidd and Huddleston (1994). In Figure 5.3, the single assessments of each question were compared with those of Kidd and Huddleston (1994). Questions 3, 4, 6, 9, and 10 were almost identical, while in question 8 the participants scored lower than that reported by Kidd and Huddleston (1994). Interestingly, in questions 1, 2, 5, and 7 the participants reported higher scores than those reported by Kidd and Huddleston (1994), and these were all associated with speed choice. The overall standard deviation of the ten questions was also slightly lower than that found in previous research (SD = 0.88 compared with SD = 1.08). Dividing the participants according to mean DPQ score resulted in two groups with 15 participants in each group. Figure 5.2 shows the resulting risk assessment of these groups, by situation. Significant differences were investigated using t-tests, which indicated that the only significant difference was found on the speed low situation ( p = 0.023; r = 0.444). The low risk group perceived this situation to be less risky than the high risk group. A reverse effect, although it did not reach significance ( p = 0.091; r = 0.314), appeared in the overtake on a bend situation. There were no other statistically significant effects.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 70

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:20:13 AM

Mean risk scores of situations Figure 5.2 9781472414694_Dorn.indb 71

5/30/2013 10:20:14 AM


Discussion

op

fC

Figure 5.3 Average DPQ-scores

oo

The results show that swerving, tailgating and risky overtaking manoeuvres are good examples of risky driving behaviour. Furthermore, tailgate medium was significantly lower than tailgate obvious ( p = 0.001; r = 0.852), swerve obvious was significantly higher than swerve medium ( p = 0.01; r = 0.485), and overtake bend significantly higher than overtake cut ( p = 0.001; r = 0.823). Surprisingly, the speed of the risky driver does not seem to be a suitable example of risky driving behaviour. While the risk score was low for all situations concerning speed choice, the variables were also not sufficiently sensitive to discriminate between situations. Thus, for example, there was no difference found between speeding obvious and speeding low ( p > 0.05; r = 0.159). This lack of a difference is despite the fact that the risky vehicle in speeding obvious exceeds the speed limit by three times as much (54 km/h) as the risky driver in speeding low (18 km/h). The standard deviations for the mean values were quite high, which may be due to subjective risk assessment and participant’s risk perception, which may vary according to previous experiences. This is especially true for the three situations in Cluster 2, which seem to be characterised by a high subjective influence. Therefore, using ratings of a risky situation may not be suitable for a large number of participants. Regarding the different DPQ groups, an interpretative approach can only be made regarding the two situations which resulted in a significant difference

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

y

72

9781472414694_Dorn.indb 72

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:00 AM


fC

References

op

y

or an observed tendency. The drivers who assessed themselves as more risky may run into risky drivers with a higher speed difference than less risky drivers would, which is why a vehicle driving too slowly may receive a higher risk score. Another explanation would be that those drivers who were more likely to offend may be more bothered by a slow car than the less risky drivers. All participants experienced the overtake bend situation in a similar way. The difference in the scores between the two groups can only be explained by the different perceptions of the group members. The higher DPQ scores, compared to Kidd and Huddleston (1994), may be due to statements 1, 2, 5 and 7, which represent excessive speed in different situations. Although the translation was executed as carefully as possible, differences due to the translation into German cannot be completely excluded. Nevertheless, the reason for differences in self-reported speeding may also be due to the different age and gender distribution within the samples, as this study was conducted mostly with younger male drivers. Support for this can be drawn from the fact that Kidd and Huddleston (1994) found the DPQ score was significantly correlated with age ( p = 0.001; r = −0.45) and gender ( p = 0.01; r = −0.16). Another cause may be the cultural and behavioural differences between German and American drivers.

Begg, D., and Langley, J. (2001). Changes in risky driving behaviour from age 21 to 26 years. Journal of Safety Research, 32(4), 491–99. Benda, H.V., and Hoyos, C.G. (1983). Estimating hazards in traffic situations. Accident Analysis and Prevention, 15(1), 1–9. Brown, I.D., and Groeger, J.A. (1988). Risk perception and decision taking during the transition between novice and experienced driver status. Ergonomics, 31(4), 585–97. Hoyos, C.G. (1988). Mental load and risk in traffic behaviour. Ergonomics, 31, 571–84. Jackson, L., Chapman, P., and Crundall, D. (2009). What happens next? Predicting other road users’ behaviour as a function of driving experience and processing time. Ergonomics, 52, 154–64. James, L., and Nahl, D. (2000). Road Rage and Aggressive Driving: Steering Clear of Highway Warfare. Amherst, NY: Prometheus. Jessor, R., Turbin, M.S., and Costa, F.M. (1997). Predicting Developmental Change in Risky Driving: The Transition to Young Adulthood. Applied Developmental Science, 1(1), 4–16. Kidd, P. (1994). Psychometric Properties of the driving practices questionnaire assessment of risky driving. Research in Nursing & Health, 17, 51–58. Knappe, B., Keinath, A., and Bengler, K. (2007). Driving Simulator as an Evaluation Tool: Assessment of the Influence of Field Of View and Secondary

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

73

9781472414694_Dorn.indb 73

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:00 AM


74

oo

fC

op

y

Tasks on Lane Keeping and Steering Performance. Retrieved from http:// 1 2 www-nrd.nhtsa.dot.gov/pdf/esv/esv20/print6.pdf, 69–79, [18.02.2013]. Lajunen, T., and Parker, D. (2001). Are aggressive people aggressive drivers? A 3 study of the relationship between self-reported general aggressiveness, driver 4 anger and aggressive driving. Accident Analysis and Prevention, 33, 243–55. 5 Minoiu Enache, N., Netto, M., Mammar, S., and Lusetti, B. (2009). Driver steering 6 assistance for lane departure avoidance. Control Engineering Practice, 17(6), 7 8 642–51. Pizza, F., Contardi, S., Mostacci, B., Mondini, S., and Cirignotta, F. (2004). A 9 driving simulation task: Correlations with Multiple Sleep Latency Test. Brain 10 11 Research Bulletin, 63(5), 423–26. Sarkar, S., Martineau, A., Emami, M., Khatib, M., and Wallace, K. (2000). Spatial 12 and temporal analyses of the variations in aggressive driving and road rage 13 behaviors observed and reported on San Diego freeways. Transportation 14 15 Research Record, 1724(1), 7–13. Shinar, D. (1998). Aggressive driving: The contribution of the drivers and the 16 situation. Transportation Research Part F: Traffic Psychology and Behaviour, 17 18 1(2), 137–60. Tasca, L. (2000, October). A review of the literature on aggressive driving research. 19 Paper presented at the Global Web Conference on Aggressive Driving Issues. 20 Taubman, O.B.A., Mikulincer, M., and Iram, A. (2004). A multi-factorial 21 framework for understanding reckless driving – appraisal indicators and 22 perceived environmental determinants. Transportation Research Part F: 23 24 Traffic Psychology and Behaviour, 7(6), 333–49.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 74

25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:00 AM

Chapter 6

1 2 3 4 5 6 7 8 9 Peter Chapman and Jodie Walton 10 University of Nottingham, UK 11 12 13 14 15 Introduction 16 In order to drive safely through potentially hazardous situations it is necessary 17 that drivers adopt an efficient visual search strategy and have an ability to detect 18 emerging dangerous situations. We suggest that drivers who are frustrated will be 19 less able to detect hazards in the road environment. Many psychological theories 20 link frustration with subsequent feelings of anger and acts of aggression. Anger 21 and aggression when driving have been shown to be associated with increased 22 accident risk. Although most authors have assumed that the link between driving 23 anger and accident risk is because of risks directly incurred through an aggressive 24 driving style, we propose that an additional reason why anger may be associated 25 with accidents is that feelings of frustration and anger directly impair drivers’ 26 27 ability to detect hazards. Driver distraction has been defined as “a diversion of attention away from 28 activities critical to safe driving towards a competing activity” (Lee, Young and 29 Regan, 2009, p. 38). Generally research on driver distraction has focused on 30 distractions that are external to the driver, be they within the vehicle, such as mobile 31 telephones (e.g., Strayer and Johnston, 2001; Rakauskas, Gugerty and Ward, 32 2004), in-car entertainment systems (Stevens and Minton, 2001) and passengers 33 (Simons-Morton, Lerner and Singer, 2005), or external to the vehicle such as 34 roadside advertising (e.g., Crundall, van Loon and Underwood, 2006; Horberry 35 and Edquist, 2009). However, there is good reason to think that internal sources 36 of distraction may be important causes of road accidents too. Possible sources of 37 internal distraction might include daydreaming (Chapman, Ismail and Underwood, 38 1999), fatigue (Connor, Whitlock, Norton and Jackson, 2001), or general emotions 39 such as grief (Rosenblatt, 2004), or anger (e.g., Lawton and Nutter, 2002; 40 Stephens and Groeger, 2011, 12; Sullman, 2006; Underwood, Chapman, Wright 41 and Crundall, 1999). Such internal distractions may be responsible for accidents 42 in much the same way that external distractions are – by diverting attention away 43 from activities critical to safe driving. The current paper considers one potential 44

fC

op

y

The Impact of Frustration on Visual Search and Hazard Sensitivity in Filmed Driving Situations

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 75

5/30/2013 10:21:00 AM


76

oo

fC

op

y

internal source of distraction – a feeling of frustration – and explores the impacts 1 that this may have on attention while responding to hazardous driving situations. 2 Frustration has traditionally been thought of as a root cause of anger and 3 aggression (Dollard, Doob, Miller, Mowrer and Sears, 1939). In the context of 4 driving this is important because frustration is a common component of everyday 5 driving. Sometimes frustration may occur simply because the traffic flow is slow 6 and large volumes of traffic are preventing the driver from obtaining their desired 7 destination. On other occasions the driver’s frustration may be caused by a particular 8 vehicle that is holding up progress by driving slowly or failing to allow the driver to 9 enter the traffic stream. In both such situations it is easy to see how frustration can 10 lead to anger, either non-specific anger, or anger at a particular road user that may be 11 followed by aggressive behaviour such as using the horn, making verbal gestures, 12 or aggressive close following. A number of studies have found that anger and 13 aggression when driving are associated with increased accident risk (Deffenbacher, 14 Deffenbacher, Richards and Lynch, 2003; Hemenway and Solnick, 1993; Parry, 15 1968; Underwood et al., 1999). Although most authors have assumed that the 16 link between driving anger and accident risk is because of risks directly incurred 17 through an aggressive driving style (e.g., King and Parker, 2008), we propose that 18 an additional reason why anger may be associated with accidents is that feelings of 19 20 frustration and anger could directly impair drivers’ ability to detect hazards. One possible mechanism by which frustration could directly impair drivers’ 21 abilities to detect hazards is through a general increase in arousal in frustrating 22 situations. For example, Otis and Ley (1993) found that frustrating participants by 23 withdrawing a reward caused them to increase the force they used in subsequent 24 behavioural responses, and was also associated with a significant increase in skin 25 conductance. Increases in arousal have been traditionally associated with a narrowing 26 in the range of cues attended to (e.g., Easterbrook, 1959). Although this can sometimes 27 be beneficial, there are cases where it means that important peripheral cues may not 28 be attended to (e.g., Loftus, Loftus, and Messo, 1987). Driving appears to be a classic 29 example of an environment where a failure to attend to peripheral cues in arousing 30 situations could cause people to fail to detect potential hazards (e.g., Chapman and 31 Groeger, 2004; Crundall, Underwood and Chapman, 2002). Thus, it has been found 32 that in dangerous or demanding situations drivers reduce their spread of visual 33 search and spend longer fixating on individual items (Chapman and Underwood, 34 1998a; Crundall and Underwood, 1998; Falkmer and Gregersen, 2005; Underwood, 35 Crundall and Chapman, 2002; Underwood, Chapman, Bowden and Crundall, 2002). 36 Although this focusing serves an important purpose in allowing the driver to process 37 risk-related central information, it has also been shown that increases in driving 38 experience allow this focusing to be reduced. Thus, Chapman and Underwood 39 (1998a) found that the increase in fixation durations in hazardous situations was 40 reduced for more experienced drivers. It appears that experience allows us to extract 41 information from the traffic scene faster and hence move on from having processed 42 an initial hazard to look for more sources of potential danger. Theoretically then we 43 might predict that frustration when driving would be associated with increased levels 44

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 76

5/30/2013 10:21:00 AM

The Impact of Frustration on Visual Search and Hazard Sensitivity

fC

op

y

of arousal (that could be measured by increases in skin conductance or heart rate) and with increases in inappropriate focusing in visual search. The current study sets out to test these predictions by inducing frustration in drivers and then exploring measures of arousal and visual search in driving related scenarios. Although small increases in arousal may be beneficial to general driving, we are particularly interested in the influences of excess arousal in situations where arousal would already be anticipated. We therefore want to look at drivers’ responses in hazardous situations to see whether additional arousal can cause them to become over-focused in such situations and cause a reduction in the driver’s ability to spot subsequent hazards. Although it would be attractive to use simulated or real driving in such a study, it is difficult to reliably induce dangerous situations in a simulator and unethical to do so on the road. Moreover, in a simulator it is possible that frustration would change the actual driving behaviour of participants. Although any such changes would be of considerable interest for future research, it is difficult to directly compare visual behaviour unless the visual stimuli remain constant between conditions. It was thus decided to use a primary task in which participants would watch hazardous driving videos while eye movements and psychophysiology could be recorded. Although we want to induce frustration in our participants we were keen to use a task in which the aim of the experiment was not apparent to the participants and where we could closely control the degree of frustration experienced. Frustration caused by blocking a goal-directed behaviour can lead directly to anger and aggression (e.g., Hanratty, O’Neal, and Sulzer, 1972) thus we were keen to find a simple task that would block participants’ goals in a way that would not be readily apparent. One such task is the unsolvable anagram task (e.g., Aspinwall and Richter, 1999). Participants generally find solving anagrams an enjoyable and motivating task, but they find it very hard to know whether an anagram is solvable or not and will continue attempting to solve impossible anagrams for extended periods if no alternatives are available. We thus chose to manipulate frustration by having participants attempt to solve anagrams that were either possible or impossible before viewing dangerous driving situations to see whether frustration induced by the anagram task would impact on visual search and the processing of hazardous driving situations.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

77

Method Design

The study used a 2 × 2 repeated measures design with two factors, Frustration (frustrated vs. control) and Hazard (dangerous vs. safer). The dependent variables were a continuous rating of risk and measures of skin conductance, heart rate, and eye movements while viewing hazard perception films. Ethical clearance for the study was obtained from the University of Nottingham’s psychology ethics committee.

9781472414694_Dorn.indb 77

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:00 AM


78

Participants The participants were 40 young drivers who were predominantly undergraduate and postgraduate students at the University of Nottingham. The sample size was chosen to provide a power of 0.87 in order to detect a medium effect size (f = 0.25) of frustration or danger on any of the dependent variables (Cohen, 1988). All participants had held a full UK driving licence for at least six months. They reported having driven an average of 16,502 miles since passing their driving tests and were aged between 20 and 33 years old, with a mean age of 21.1 years. The sample consisted of 17 males and 23 females.

y

Stimuli

oo

fC

op

The critical stimuli consisted of a series of eight hazard perception clips provided by the UK Driving Standards Agency (DSA). These clips are licensed by the DSA for use in training and research and are similar to those used in the hazard perception component of the UK driving theory test. Each clip lasted approximately one minute and contained a driver’s eye view of a vehicle negotiating a potentially hazardous driving situation. Examples of hazards included pedestrians stepping out into the road, cyclists crossing into the driver’s path, cars pulling out suddenly from the side of the road, and motorcyclists failing to give way to the oncoming vehicle. Hazards had been selected to have a gradual onset and to have the potential to be predicted in advance by an experienced driver. Individual clips contained between one and three hazards and each hazard was marked from the moment the potential hazard first became visible to the driver to the moment the hazard left the screen. For analysis purposes (following Chapman and Underwood, 1998a) this allowed us to split each clip into sections where a hazard was present (dangerous) and the remaining sections where no hazard was in progress (safer). Three additional hazard perception clips of the same type were used for a practice session. Frustration was manipulated using a series of 16 anagrams. These consisted of a series of shuffled letter strings of between five and eight characters in length. Half of these letter strings were solvable anagrams in that they could be rearranged into a standard English word (e.g., RCHIA becomes CHAIR) while the other half were unsolvable (e.g., WEMOL). The final 16 anagrams were selected from a longer list that was trialled by three pilot participants. In this pilot task participants were asked to rate each anagram for solvability (without actually attempting to solve it). The eight unsolvable anagrams were all reliably rated as solvable in the pilot task. An additional six solvable anagrams were used for the practice block. Hazard perception clips and anagrams were edited onto miniDV digital video tapes using Final Cut Pro. A pair of anagrams appeared on the screen for 40 seconds, followed by a hazard perception clip, followed by a series of five questions about each clip (i.e., how dangerous was the clip, how tired did the participant feel, how frustrated did they feel, how fast did they feel the vehicle was going in the clip, and how hard was it to spot the hazard). Each tape contained

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 78

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:00 AM


fC

Procedure

op

y

11 repetitions of this sequence of anagram pair, hazard clip, five questions. The first three repetitions of this sequence served as a practice session, were the same for all participants and included only solvable anagrams. The remaining eight sequences were blocked into an ABBA design, such that participants would watch half the clips after solvable anagrams (control condition) and half after unsolvable anagrams (frustration condition). To counterbalance the link between condition and hazard clip and the order of clip presentation, four separate stimulus tapes were created. There was no sound track on the video, instead audio markers were added to the tape to mark the starts and ends of sections. These markers were used for analysis purposes and were not audible to participants watching the videos. To avoid participants realising that the anagram blocks were designed to influence performance on the driving, there was no anagram trial before the first hazard perception video, and one was added after the final video. This meant that from the participant’s perspective they responded to a hazard perception video, then answered questions on it, then attempted to solve anagrams. However for analysis purposes we have blocked the previous set of anagrams with the subsequent video to measure the effect of frustration from the anagrams on subsequent hazard perception performance.

Participants had the study explained to them after which they completed an informed consent form. They then sat down one metre away from a dual computer/ video monitor subtending approximately 21° of visual angle horizontally and 16° of visual angle vertically. A Biopac MP150 biopotential amplifier was used to record psychophysiology and responses from the participant and to link these with the audio markers on the tapes. Skin conductance was recorded using a GSR100c preamplifier connected to TSD203 skin conductance transducers which were filled with GEL101 isotonic electrode paste and attached using Velcro straps to the intermediate phalange of the first and third fingers of the participant’s nondominant hand. Pulse was recorded using a PPG100c photo-plethysmograph preamplifier connected to a TSD200 transducer which was placed on the distal phalange of the participant’s second finger from their non-dominant hand. The participant used their dominant hand to make a continuous hazard rating using a Biopac variable response transducer (TSD115) attached to the high level transducer interface (HLT100c) of the MP150. This allowed participants to rate danger continuously on a 10 point scale from Very safe to Very dangerous while watching the hazard perception clips. Audio tracks from the video were also fed into the MP150 directly via the UIM100c analogue channel input. This allowed auditory markers from the video to be recorded on the same time-scale as the continuous responses and psychophysiology. Once the psychophysiology was set up and found to be working, participants were then calibrated on an SMI remote eye-tracking device (RED). This allowed free head recording of gaze location at a sample rate of 50hz and allowed head

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

79

9781472414694_Dorn.indb 79

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:00 AM


80

movements within an area of about 20 cm in each direction. A chin rest was used to limit head movements to well within this range. After a nine-point calibration routine the experiment started. Participants viewed a hazard perception clip while making a continuous rating of danger, then answered five questions about the clip, and then had 40 seconds to solve two anagrams. Participants’ responses to the questions and anagrams were spoken out loud and recorded by the experimenter. After all 11 videos had been viewed the participant was thanked for their participation and debriefed fully. Results

y

Anagrams and ratings

oo

fC

op

In the experimental conditions participants had to attempt to solve eight pairs of anagrams, four of which were solvable, and four of which were unsolvable (although the participants did not know this). On the solvable trials participants correctly solved both anagrams in the pair on 76 per cent of occasions, only one of the pair on 17 per cent of occasions, and neither of them on the remaining 7 per cent of occasions. Participants never succeeded in solving either item from the unsolvable anagram pairs. Participants found the anagram task to be engaging and at debrief they generally claimed to be annoyed at not having been able to solve all the pairs. No participant reported realising that some of the anagrams were actually impossible to solve. The eight key hazard clips were divided for each participant into those that followed unsolvable anagrams (frustrated condition) and those that followed solvable anagrams (control condition). Participants rated hazard clips following unsolvable anagrams as marginally more dangerous (M = 3.25) than those following solvable anagrams (M = 3.08, t(39) = 1.66, p = 0.10), but otherwise there were no differences in post-clip ratings between the two conditions. Critically the participants did not rate themselves as more frustrated when answering questions about hazard perception videos that followed unsolvable anagrams (M = 2.18) than those that followed solvable anagrams (M = 2.23, t(38) = −0.56, p = 0.57). We interpret this to mean that while frustration from unsolvable anagrams may have affected the way in which drivers viewed the hazard perception clips, they were not explicitly aware of their frustration by the time each hazard perception clip had finished.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Continuous hazard ratings For all remaining analyses we have continuous data from throughout each hazard video, so clips were divided into dangerous and safer sections and analyses are reported of the 2 × 2 repeated measures ANOVAs with Hazard and Frustration as

9781472414694_Dorn.indb 80

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:00 AM

op

fC

Figure 6.1 Ratings and skin conductance measures during dangerous and safer sections of hazard videos divided by whether the participant had previously been frustrated by attempting to solve impossible anagrams or not (control condition) factors. Where significant interactions were found we report analyses of simple main effects to explore the interaction. Figure 6.1a shows the mean danger level given using the variable response transducer in frustrated and control conditions. There was a main effect of Hazard (F(1, 39) = 90.56, p < 0.001), with dangerous sections receiving higher mean ratings than safer ones, and a marginal interaction between Hazard and Frustration (F(1,39) = 3.26, p = 0.079). Simple main effects analysis confirmed that in dangerous sections frustrated drivers gave marginally lower ratings than the control drivers (F(1,39) = 3.80, p = 0.059), while there was no difference in ratings for the safer sections (F(1,39) < 0.01, p = 0.983). To gain a measure of variability in hazard ratings throughout the clip sections the standard deviation of response lever position was calculated for each clip type and this is plotted in Figure 6.1b. Again, there was a significant main effect of hazard (F(1,39) = 88.92, p < 0.001), with dangerous sections receiving more variable responses. There was also a significant interaction between Hazard and Frustration (F(1,39) = 8.13, p = 0.007). Simple main effects analysis confirmed that frustrated drivers made less variable responses in dangerous situations

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

81

y


9781472414694_Dorn.indb 81

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:01 AM


82

(F(1,39) = 8.01, p = 0.007), while there was no significant difference in the safer 1 2 situations (F(1,39) = 1.16, p = 0.288). Psychophysiological measures

oo

fC

op

y

Mean skin conductance for each section was calculated in microSiemens (µS). A marginal main effect of Frustration was observed (F(1,39) = 3.49, p = 0.069), with frustrated drivers having higher skin conductance than when they were in the control condition. This is shown in Figure 6.1c, however, there were no significant effects of Hazard, or interactions between Hazard and Frustration on this measure. To get a measure of variability of skin conductance the standard deviation of skin conductance was calculated within each clip section. These data are plotted in Figure 6.1d. There was a main effect of hazard with skin conductance variability being lower in the dangerous sections (F(1,39) = 40.52, p < 0.001). There was also a significant main effect of frustration with skin conductance being more variable for frustrated drivers than when they were in the control conditions (F(1,39) = 6.29, p = 0.017). There was also an interaction between Frustration and Hazard (F(1,39) = 6.57, p = 0.014). Simple main effects analyses revealed that frustrated drivers had more variability in skin conductance in safer situations (F(1,39) = 7.71, p = 0.008), but not in the dangerous ones (F(1,39) = 0.59, p = 0.448). As a final measure of skin conductance, individual electrodermal responses (EDRs) were scored by eye and counted in each section of the video. To account for differences in section length these were then converted into an overall measure of electrodermal responses per minute. These data are plotted in Figure 6.2a and show a significant interaction between Frustration and Hazard (F(1,39) = 9.66, p = 0.004). Simple main effects analysis revealed that frustrated drivers made fewer electrodermal responses per minute than when in the control condition, but only while watching dangerous video sections (F(1,39) = 4.99, p = 0.031). No difference as a function of frustration was present in the safer video sections (F(1,39) = 2.80, p = 0.102). Heart rate was derived from the pulse signal using the automatic rate calculations in AcqKnowledge software and is shown in beats per minute (bpm) in Figure 6.2b. There was a main effect of Hazard, with heart rate generally being lower in the dangerous sections (F(1,39) = 42.61, p < 0.001), and an interaction between Frustration and Hazard (F(1,39) = 8.80, p = 0.005). Analysis of simple main effects revealed that heart rate was marginally higher in frustrated drivers during the safer sections (F(1,39) = 3.96, p = 0.054), but not during the dangerous sections (F(1,39) = 1.64, p = 0.209).

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Eye movement measures Fixations were calculated from the raw eye movement data using an IDF Converter with a spatial dispersion threshold of 50 pixels (approximately 1° of visual angle) and a minimum fixation duration set at 80 ms. For three participants the

9781472414694_Dorn.indb 82

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:02 AM


y op

Physiological and eye movement measures during dangerous and safer sections of hazard videos divided by whether the participant had previously been frustrated by attempting to solve impossible anagrams or not (control condition)

fC

Figure 6.2

eye-tracker failed to calibrate satisfactorily, therefore eye movement measures are reported for 37 drivers only. Figure 6.2c shows the mean fixation durations during each clip section. There was a main effect of Frustration, with mean fixation durations tending to be shorter for frustrated drivers (F(1,36) = 5.27, p = 0.028). There was also a main effect of Hazard (F(1,36) = 13.78, p < 0.001), with fixation durations tending to be longer when watching dangerous clip sections. There was no significant interaction between Frustration and Hazard. To gain an overall measure of spread of search the standard deviation of fixation locations was calculated for each clip section separately in the horizontal meridian and the vertical meridian. Spread of horizontal search is plotted in Figure 6.2d. There was a main effect of Hazard with spread of search being significantly lower during dangerous clip sections (F(1,36) = 28.33, p < 0.001), but no main effect of Frustration or interaction between Frustration and Hazard. Spread of vertical search showed a similar pattern, with a main effect of Hazard (F(1,36) = 69.86, p < 0.001), with lower spread of vertical search in dangerous clip segments, and again no main effect of Frustration or interaction between Frustration and Hazard.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

83

9781472414694_Dorn.indb 83

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM


84

Discussion

oo

fC

op

y

The initial analysis of anagram performance confirmed that participants were almost always able to solve at least one anagram in the control conditions and usually both, while they were never able to solve anagrams in the frustrated condition. Despite these differences in performance there were no significant differences when they later rated their feelings of frustration after watching a hazard perception video. One possibility is that the drivers’ ratings were linked to the frustration inherent in the clip they had just watched and were not sensitive to the internal frustration carrying over from the anagram task. Another possibility is that any frustration effect had worn off by the time they came to do each rating task. However, there was plenty of evidence that the frustration manipulation did affect their performance on the other tasks, so our conclusion is that the frustration manipulation was successful, but that the participants were not explicitly aware of their frustration a minute later after watching a hazard perception clip. The fact that clips were overall rated as marginally more dangerous after a frustration manipulation is interesting, particularly when it is contrasted with the continuous manual ratings of hazard that were provided when viewing the clip. Here we found a marginal tendency for drivers to rate clips as less dangerous after the frustration manipulation, though only during the dangerous sections of the clips. Similarly, drivers gave less variance in danger ratings when they had been frustrated and were watching dangerous video sections. These findings suggest that after the frustration manipulation drivers were less sensitive to the pre-planned hazards in the dangerous video sections, though they may subsequently have increased their overall danger ratings partly because they failed to spot the critical hazards at the time. The interpretation that frustrated drivers were less aware of the pre-planned hazards is consistent with the data from the psychophysiology. Here frustration led, as predicted from Otis and Ley (1993), to generally higher skin conductance. It also led to more variance in skin conductance during the safer clip sections, however, drivers actually produced fewer electrodermal responses per minute during dangerous clip sections after the frustration manipulation. This seems consistent with the idea the excessive arousal is actually making participants less sensitive during the critical hazards. Dangerous video sections were generally associated with heart rate slowing, although the degree to which this occurred did not differ as a function of frustration. Heart rate slowing is often linked to the occurrence of unexpected events (e.g., Somsen, van der Molen, Jennings and van Beek, 2000) and is a plausible response to the occurrence of unexpected hazards. Like mean skin conductance, however, there was a general tendency for heart rate to be higher in the safer sections after the frustration manipulation. Overall, the psychophysiological measures paint a picture of frustration being associated with a generally higher level of arousal and an increased sensitivity to hazards during safer sections of the videos. In contrast it would appear that frustrated drivers are actually less sensitive to the serious hazards that occur during the dangerous sections of the clips.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 84

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM


The eye movement measures show a pattern of results that is highly consistent with previous studies in which drivers have viewed hazard perception clips (e.g., Chapman and Underwood, 1998a, 1998b). Generally fixation durations increase in dangerous situations, and the spread of search, both vertically and horizontally, decreases in such situations. This is consistent with a pattern of attention focusing in hazardous situations. We have previously interpreted this increase in fixation durations in hazardous situations as evidence for deeper processing of hazardrelated information. The fact that this increased fixation duration on hazard-related information is reduced by experience or training (Chapman and Underwood, 1998a; Chapman, Underwood and Roberts, 2002), is consistent with the idea that this is time spent extracting information from potential sources of hazard in the environment. In this context it is interesting to note that frustration was associated with relatively short fixation durations across all types of driving situation. The most likely interpretation is that frustration is causing shallower processing of visual information from the scene. In the safer sections this may account for greater variability in skin conductance, while in the dangerous situations this could account for a failure to be sufficiently sensitive to hazards both when measured by continuous ratings and by the number of electrodermal responses produced. In conclusion, the current study provides evidence that frustration caused by an irrelevant task can lead to increases in arousal that carry over into a driving task. Although this increased arousal is not problematic during easy driving situations (and it may even be beneficial), it does create a problem during hazardous situations. Here drivers are less sensitive to hazards, both behaviourally and psychophysiologically. Eye movement recording suggest a mechanism by which this may be happening. Frustration and increased arousal appear to be associated with a reduction in fixation durations that is consistent across all driving situations. It may be that arousal is causing drivers to disengage their attention too soon from the items they are fixating (cf. Findlay and Walker, 1999) and to attend to alternative distractions in the driving scene before they have fully processed all relevant information. This may make them insensitive to some relevant hazards and provide a possible alternative mechanism for increases in accident risk associated with angry and aggressive drivers. An exciting next step for this research would be to move it into a simulated driving environment to explore interactions between frustration and actual driving behaviour, and to look for direct mechanisms for accident causation.

fC

op

y

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 References 39 Aspinwall, L.G., and Richter, L. (1999). Optimism and self-mastery predict more 40 rapid disengagement from unsolvable tasks in the presence of alternatives. 41 42 Motivation and Emotion, 23, 221–45. Chapman, P., and Underwood, G. (1998a). Visual search of driving situations: 43 44 Danger and experience. Perception, 27, 951–64.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

85

9781472414694_Dorn.indb 85

5/30/2013 10:21:03 AM


86

oo

fC

op

y

Chapman, P., and Underwood, G. (1998b). Visual search of dynamic scenes: Event types and the role of experience in viewing driving situations. In G. Underwood (ed.), Eye guidance in reading and scene perception, (pp. 371–96). Oxford: Elsevier. Chapman, P., Ismail, R., and Underwood, G. (1999). Waking up at the wheel: Accidents, attention and the time-gap experience. In A.G. Gale, I.D. Brown, S.P. Taylor and C.M. Haslegrave (eds), Vision in Vehicles 7 (pp. 131–38). Oxford: Elsevier. Chapman, P., Underwood, G., and Roberts, K. (2002). Visual search patterns in trained and untrained novice drivers. Transportation Research Part F, 5, 157–67. Chapman, P., and Groeger, J.A. (2004). Risk and the recognition of driving situations. Applied Cognitive Psychology, 18, 1–19. Cohen, J. (1988). Statistical power analysis for the behavioural sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum & Associates. Connor, J., Whitlock, G., Norton, R., and Jackson, R. (2001). The role of driver sleepiness in car crashes: A systematic review of epidemiological studies. Accident Analysis and Prevention, 33, 31–41. Crundall, D.E., and Underwood, G. (1998). The effects of experience and processing demands on visual information acquisition in drivers. Ergonomics, 41, 448–58. Crundall, D., Underwood, G., and Chapman, P. (2002). Attending to the peripheral world while driving. Applied Cognitive Psychology, 16, 459–75. Crundall, D., van Loon, E., and Underwood, G. (2006). Attraction and distraction of attention with roadside advertisements. Accident Analysis and Prevention, 38, 671–77. Deffenbacher, J., Deffenbacher, D., Richards, T., and Lynch, R. (2003). Anger, aggression and risky behavior: A comparison of high and low anger drivers. Behavior Research and Therapy, 41, 701–18. Dollard, J., Doob, L., Miller, N., Mowrer, O., and Sears, R. (1939). Frustration and aggression. New Haven: Yale University Press. Easterbrook, J. (1959). The effect of emotion on cue utilization and the organization of behavior. Psychological Review, 66, 183–01. Falkmer, T., and Gregersen, N. (2005). A comparison of eye movement behaviour of inexperienced and experienced drivers in real traffic environments. Optometry and Vision Science, 82, 732–39. Findlay, J., and Walker, R. (1999). A model of saccadic eye movement generation based on parallel processing and competitive inhibition. Behavioral and Brain Sciences, 22, 661–721. Hanratty, M., O’Neal, E., and Sulzer, J. (1972). Effect of frustration upon imitation of aggression. Journal of Personality and Social Psychology, 21, 30–34. Hemenway, D., and Solnick, S. (1993). Fuzzy dice, dream cars, and indecent gestures: Correlates of driver behavior? Accident Analysis and Prevention, 25, 161–70.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 86

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM


fC

op

y

Horberry, T., and Edquist, J. (2009). Distractions outside the vehicle. In M. Regan, J. Lee and K. Young (eds), Driver distraction: Theory, effects, and mitigation (pp. 215–27). Boca Raton, FL: CRC Press. King, Y., and Parker, D. (2008). Driving violations, aggression and perceived consensus. European Review of Applied Psychology, 58, 43–49. Lawton, R., and Nutter, A. (2002). A comparison of reported levels and expression of anger in everyday and driving situations. British Journal of Psychology, 93, 407–23. Lee, J.D., Young, K.L., and Regan, M.A. (2009). Defining driver distraction. In M. Regan, J. Lee and K. Young (eds), Driver distraction: Theory, effects, and mitigation (pp. 31–40). Boca Raton, FL: CRC Press. Loftus, E.F., Loftus, G.R., and Messo, J. (1987). Some facts about “weapon focus”. Law and Human Behavior, 11, 55–62. Otis, J., and Ley, R. (1993). The effects of frustration induced by discontinuation of reinforcement on force of response and magnitude of the skin conductance response. Bulletin of the Psychonomic Society, 31, 97–100. Parry, M. (1968). Aggression on the road. London: Tavistock. Rakauskas, M.E., Gugerty, L.J., and Ward, N.J. (2004). Effects of naturalistic cell phone conversations on driving performance. Journal of Safety Research, 35, 453–64. Rosenblatt, P. (2004). Grieving while driving. Death Studies, 28, 679–86. Simons-Morton, B., Lerner, N., and Singer, J. (2005). The observed effects of teenage passengers on the risky driving behavior of teenage drivers. Accident Analysis and Prevention, 37, 973–82. Somsen, R., van der Molen, M., Jennings, R., and van Beek, B. (2000). Wisconsin card sorting in adolescents: Analysis of performance, response times, and heart rate. Acta Psychologica, 104, 227–57. Stephens, A.N., and Groeger, J.A. (2011). Anger congruent behaviour transfers across driving situations. Cognition and Emotion, 25(8), 1423–38. Stephens, A.N., and Groeger, J.A. (2012). Driven by anger: The causes and consequences of anger during virtual journeys. In M. Sullman and L. Dorn (eds), Advances in traffic psychology (pp. 3–15). Farnham, UK: Ashgate. Stevens, A., and Minton, R. (2001). In-vehicle distraction and fatal accidents in England and Wales. Accident Analysis and Prevention, 33, 539–45. Strayer, D.L., and Johnston, W.A. (2001). Driven to distraction: Dual-task studies of simulated driving and conversing on a cellular phone. Psychological Science, 12, 462–66. Sullman, M.J.M. (2006). Anger amongst New Zealand drivers. Transportation Research Part F, 9, 173–83. Underwood, G., Chapman, P., Wright, S., and Crundall, D. (1999). Anger while driving. Transportation Research Part F, 2, 55–68. Underwood, G., Chapman, P., Bowden, K., and Crundall, D. (2002). Visual search while driving: Skill and awareness during inspection of the scene. Transportation Research Part F, 5, 87–97.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

87

9781472414694_Dorn.indb 87

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM


88

Pr

oo

fC

op

y

1 Underwood, G., Crundall, D., and Chapman, P. (2002). Selective searching while 2 driving: The role of experience in hazard detection and general surveillance. 3 Ergonomics, 45, 1–12 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 88

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM

Chapter 7

Anger and Prospective Memory While Driving: Do Future Intentions Affect Current Anger? Amanda N. Stephens*, Gillian Murphy* and Steven L. Trawley**

Introduction

y

School of Applied Psychology, ** School of Medicine, Centre for Gerontology Research, University College Cork, Ireland

op

*

Anger and driving

fC

It is well established that experiences of anger have a detrimental cost on driver performance. However, less is known about how anger may interact with secondary tasks and whether this curtails or enhances the detriment to driving. In this chapter, we present preliminary data investigating how anger over being impeded by slower lead vehicles and the requirement to remember to do something in the near future (prospective memory) interact and influence behaviour during a simulated driving scenario.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

The effects of anger extend beyond a direct association between anger and aggression and can serve as a source of preoccupation (Rusting and NolenHoeksema, 1998), misdirected attention (Forgas, 1995) and distraction (Lansdown and Stephens, 2013). When examined in a driving context, simulator-based studies have shown that drivers manipulated into an angry mood appear to make more stereotypical assessments of driving situations (Stephens and Groeger, 2011) and take longer to respond to hazardous driving events (Stephens, Madigan, Trawley and Groeger, 2013). Survey-based studies have also found reliable relationships between self-reported anger and tendencies to lose concentration, make small errors and commit driving violations (Berdoulat, Vavassort and Munoz Sastre, 2013). This may suggest a tendency for angry drivers to ruminate on sources or feelings of anger, to the detriment of the driving task. Overall, when we have manipulated anger in driving scenarios and examined the consequences, we have tended to find less evidence of general behaviour changes (such as increased, more erratic speeds) and more situation-specific costs (e.g., Stephens and Groeger, 2011). For example, we recently examined the eye-movement behaviour of angry drivers driving a simulated scenario and found that although drivers in an angry

9781472414694_Dorn.indb 89

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM


90

oo

fC

op

y

mood had no differences in latency to first fixation on potential hazards, they took longer to return their attention to these when there was some ambiguity about the outcome (Stephens et al., 2013). Of further concern, is that the consequences of anger can transfer into unrelated situations, which is why anger and subsequent aggression may appear unrelated and unprovoked (Stephens and Groeger, 2011). While a substantial amount of research now exists outlining the various causes and consequences of anger while driving, less has been done to consider how to reduce this anger and alleviate the detriment to driving performance. In the broader context of anger research, laboratory studies have shown that distraction reduces anger and angry mood (Bushman, 2002; Gerin, Davidson, Christenfeld, Goyl and Swartz, 2006; Rusting and Nolen-Hoeksema, 1998). Traditionally, the effect on anger of either distraction (asking the participant to think about an unrelated topic) or rumination (asking the participant to consider the source of anger) is compared. While rumination increases and/or prolongs anger, distraction reduces self-reported anger (Bushman, 2002; Rusting and Nolen-Hoeksema, 1998) and improves blood pressure recovery after arousal (Gerin et al., 2006). Moreover, when manipulated into an angry mood, women seek distraction to regulate mood (Rusting and Nolen-Hoeksema, 1998). Thus, it is not just that the increased cognitive load reduces levels of anger per se, but that the unrelated nature of the cognitive task serves to reduce anger levels. While the aim in driving would not be to replace the detrimental effects of anger with other forms of dangerous distraction, there may be suitable everyday tasks that have a beneficial effect on driver anger. In this study we explore the role of Prospective Memory (PM), defined as remembering to do something in the future, as such a distraction task, and investigate how PM tasks interact with driver anger and effect driver behaviour. Prospective memory and driving

Prospective memory is common in everyday life as we regularly maintain several concurrent intentions to do something in the near future. For example, remembering to stop at the bank on your way home from work or remembering to avoid the toll-roads on a certain route. Traditionally PM is conceptualised into three distinct phases, which are encoding, maintenance and retrieval (Einstein and McDaniel, 2005). The maintenance phase is the period between forming/ encoding an intention and being in the appropriate time or place for retrieving it. The retrieval phase is where the intention is retrieved and implemented. The majority of PM research is concerned with the retrieval phase, where a disruptive effect of PM retrieval is often found on concurrent task performance (e.g., Einstein, McDaniel, Manzi, Cochran and Baker, 2000; Rendell, McDaniel, Forbes and Einstein, 2007; Trawley, Rendell, Groeger and Stephens, under review). In contrast, the maintenance period is generally not considered to be demanding enough to interfere with task performance under standard PM task conditions (see Brandamonte, Ferrante and Delbello, 2001). However, some researchers argue that

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 90

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM

Anger and Prospective Memory While Driving

fC

op

y

maintaining the intention is moderately demanding on attentional resources when participants intermittently activate their intentions (Einstein, McDaniel, Willford, Pagan and Dismukes, 2003). For PM to be a useful paradigm to reduce anger in driving we expect it to shift attention from the source of the anger, but not to interfere with the driving task during the maintenance phase. Two studies published by Oron-Gilad, Ronen and Shinar (2008) have shown it is possible to engage drivers in some cognitive tasks without impairing driving performance. These researchers conducted simulatorbased studies, designed to increase driving-task engagement by combating fatigue (boredom). They found that fatigue could be overcome by listening to music or by participation in a trivia game and neither of these had a cost to driving performance. To date, little research has been done to indicate whether in a driving context, simply maintaining a PM intention will have a cost to performance. This is despite the fact that most drivers will frequently be maintaining such intentions while driving. The aims of the study reported in the succeeding text were to firstly investigate the influence of PM on manipulated anger. We expected that self-reported anger would be lower for drivers given a PM errand to complete. We further aimed to examine the cost, if any, to driving of both maintaining and then retrieving a PM errand. For the purposes of this study we prefer the term implementation rather than retrieval, because it places the emphasis on task performance, rather than retrieval of what has to be done. In this regard, we hypothesised that behaviour during the maintenance phase would not differ according to whether drivers had a PM errand or not. However, during the implementation phase, drivers with a PM errand would exhibit disrupted driving behaviour compared to drivers without a PM errand.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

91

Method

Participants

Twenty-eight licensed drivers affiliated with the University College Cork, Ireland were recruited for the study. One driver was removed from the analysis due to driving at excessive speeds. This resulted in 27 participants (Males = 13) who had an average age of 24 years (SD = 5.16), had been licensed for an average of 4.3 years (SD = 4.31) and drove approximately 120 (SD = 94.5) kilometres (km) per week or 5,853 (SD = 4,971) km per year. Participants were allocated into one of four groups combining PM errand (with/without) with anger-provocation (provoking impediment/un-provoking no impediment). Group allocation was predetermined before participation and based on recruitment order. The groups were statistically similar on age and driving experience (years licensed, weekly or annual mileage). Participants received €10 compensation for their time.

9781472414694_Dorn.indb 91

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM


92

University College Cork driving simulator Driving data were gathered in the STISIM 400W driving simulator belonging to the School of Applied Psychology, UCC. The simulator is a fixed-base Volkswagon Polo with manual transmission. It has 7.1 Dolby surround sound and a 135-degree field of view, resulting from image projection onto three wall-tofloor screens located approximately one to one-and-a-half metres from the car body. The simulator is equipped with side mirrors and rear projection allowing drivers the ability to monitor rear world traffic. Design and procedure

fC

op

y

All participants were required to attend one session in the driving laboratory where they provided data for two separate studies. The study reported here was the first and took approximately 30–40 minutes. Upon arrival, participants were given a practice trial in the simulator (10 minutes), which served as a screen for motion sickness. Immediately after this, informed consent was obtained from participants who were comfortable to continue with the study. Participants then provided demographic information (age, driving history) (5 minutes) and completed the Profile of Mood States – Short Bilingual Version (POMS SBV; Cheung and Lam, 2005) (5 minutes). The SmartEYETM eyetracker was also calibrated at this time, however eye-movement behaviour recordings will not be discussed in this chapter. Driving simulator task

oo

Prior to commencing the simulated drive, all participants were instructed to drive as they usually would, complying with all traffic signs and posted speed limits. They were also informed that during the drive they would be required to rate their current levels of anger. These ratings were on a five point scale (1 = Not angry to 5 = Very angry) and would be prompted by the sound of a bell. An example was provided. Data for this study were collected in one continuous 25-minute simulated drive. However, the drive could be conceptually broken down into four phases (see Table 7.1). Phase 1 lasted for approximately five minutes and consisted of a residential area where drivers were required to travel behind a lead vehicle for about one kilometre and were then allowed to travel uninterrupted for a further one kilometre. This section provided baseline information on behaviours while following a lead vehicle (headway or speed). As the study was a 2 × 2 (PM errand × anger-provocation) design, half of the drivers were impeded by the lead vehicle (anger-provocation group), which travelled far slower than the posted speed limit (averaging 20 km/h in a 50km/h zone). At the end of the baseline phase, the simulation was paused and half of the drivers were given additional PM errands. These drivers were instructed: “When you reach a street of shops you are to look out for a Bank of Ireland and a Waterstones bookstore. Every time you see one of these shops, sound your horn.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 92

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM

9781472414694_Dorn.indb 93

5/30/2013 10:21:03 AM

Phase 1: Baseline

✘

✘

✔

✔

Group 2: PM only (n = 7)

Group 3: Anger-provocation only (n = 8)

Group 4: PM and Anger-provocation (n = 6)

Impediment by lead vehicle

Group 1: Control (n = 6)

Table 7.1 Study design

✘

✘

✘

✘

PM errands

PM errands provided

✔

✔

✔

✘

✔

✘

✘

✘

PM errands


Phase 2: Maintenance

✔

✔

✔

✔

Variable speed limit

✔

✘

✔

✘

PM errands

Phase 3: Implementation

✔

✔

✘

✘


✘

✘

✘

✘

PM errands

Phase 4 Post Implementation


94

oo

fC

op

y

You don’t need to slow down to do so”. Drivers not receiving a PM task were instructed that there was a group of shops coming up soon. Phase 2 was a replication of the baseline environment and consisted of a second following task matching the parameters of the baseline follow task. For example, drivers in the anger-provocation groups, and whom the first lead vehicle impeded, were also impeded during phase 2. Whilst Phase 2 matched Phase 1 in the driving task demand, half of the drivers had the additional PM task load during this Phase. Thus, we refer to Phase 2 as the maintenance phase as drivers had their PM intentions, but were not able to act upon them until they entered the third Phase. The third phase of the drive lasted approximately 10 minutes and consisted of one long street of shops. Drivers assigned to the PM errand groups were required to implement their intentions during this section. For those drivers, there were four chances to sound their horn, as each target shop appeared twice in the simulation. Shops were strategically positioned to be more than two minutes apart and the order of these counterbalanced between PM participants. For all drivers, the implementation phase contained variable speed limits, with sign posted speeds altering between 40, 50, 60 and 70 km/h. There were 18 required changes during this section, which averaged out to a speed of 60 km/h. It should be noted that unbeknown to the drivers with the PM errands, the target shops only appeared in the 60 km/h zones. The final phase of the drive was again a residential area and contained a final lead vehicle following task and a section of unimpeded driving. Verbal anger ratings were also requested at six points during the drive. Upon completion of the drive, memory recall was tested for PM errand and drivers then completed a post-drive POMS. Results and Discussion

The results will be presented across three sections that address each specific hypothesis. First, that there will be differences in reported anger between those with PM errands and those without PM errands. Second, that maintaining a PM errand will have no cost on driver behaviour. Third, that there will be a cost to driver performance when drivers implement their PM errands.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

H1: Self-reported anger will be lower for drivers given a PM errand to complete A manipulation check of self-reported anger after the enforced follow tasks revealed that drivers in the anger-provocation groups reported reliably higher levels of anger (M = 2.77, SD = 1.10) than non-provoked drivers (M = 1.92, SD = 1.00; t (25) = −2.17, p < 0.05). To simplify the analysis and because we were mainly interested in anger during the PM maintenance and implementation phases we conducted separate 2 × 2

9781472414694_Dorn.indb 94

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:03 AM

op

fC

Figure 7.1 Self-reported anger across baseline, maintenance and implementation phases between subjects ANOVAs on self-reported anger after each phase. The between subjects variables were anger-provocation (impeded by lead driver/unimpeded by lead driver) and PM errands (with/without). These analyses allowed us to examine whether anger from being impeded was reduced for drivers who also had a secondary task to think about and implement. The ANOVA on anger after maintenance showed a main effect of anger-provocation (F(1,23) = 6.43, p < 0.01) with drivers being slowed by the lead vehicle reporting reliably higher anger (M = 3.00, SD = 1.15) than drivers unimpeded by the lead vehicle (M = 2.00, SD = 0.90). However, maintaining a PM intention had no effect on the anger levels in either anger-provocation group (see Figure 7.1). Thus, we cannot reject the null hypothesis regarding a PM and anger interaction during the PM maintenance phase. When we conducted the ANOVA on anger after the implementation phase, we found no evidence to support the hypothesis that performing a PM errand reduced anger levels beyond having no errand. No main effects of anger-provocation or PM errand and no two-way interactions were found. However, the direction of the means (outlined in Figure 7.1) indicate higher anger levels for drivers in the nonprovocation group and with a PM errand. To explore this further, we ran paired t-tests comparing anger reported after the maintenance and implementation phases for each group. These showed that anger over being impeded by a slower moving vehicle reliably decreased across the PM implementation phase for drivers with an errand (t (5) = 2.92, p < 0.05) and those without (t (7) = 7.78, p < 0.001). The secondary task therefore, neither enhanced nor decreased anger more than driving

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

95

y


9781472414694_Dorn.indb 95

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:04 AM


96

uninterrupted for a 10-minute period. However, anger levels were maintained for the drivers in the PM only group. Thus, the preliminary data suggest minor annoyances were maintained by having to perform a secondary task, whereas moderate levels of anger decrease over a 10 minute period, regardless of whether drivers had an additional task to perform or not. H2: Driving behaviour during the maintenance phase will not differ according to whether drivers do or do not have a PM task

oo

fC

op

y

Headway (range) and speed have been previously found to differ in angerprovocation groups exposed to slower-lead drivers (Stephens and Groeger, 2011) and so we used these variables to operationalise behaviour during the maintenance phase. Given that driver speed was constrained for half of the drivers during the follow tasks, we analysed following behaviours for the anger-provocation and no provocation groups separately. To understand whether the headway of the anger-provocation group altered according to whether they had a PM errand, we ran a mixed ANOVA with a within subjects factor of phase (baseline/maintenance) and between groups factor of PM errands (with/without) for both average and variation of headway (metres). We found little evidence of an effect of PM on driving behaviour.1 For average range, there was no main effect of phase and no interaction between phase and PM group. When we examined the variability in range, a main effect of phase was also not significant and no interaction found. However, a reliable between groups difference emerged (F(1,9) = 7.03, p < 0.05). Although the interaction was not significant, the between groups difference indicated that groups may have differed at the baseline drive and so we performed follow up independent t-tests to examine group differences in headway variability, first at baseline and then during the maintenance phase. Independent t-tests on the variation of range during the baseline follow task, showed no mean differences between the two groups. However, drivers with a PM errand had more variation in the headway they allowed between themselves and the lead vehicle during the maintenance phase (M = 4.73, SD = 1.95), when compared to anger-provoked drivers with no PM errand (M = 2.35, SD = 0.92; t(11) = −2.41, p < 0.05). Thus, providing tentative evidence that drivers maintaining a PM intention had to work harder to maintain appropriate distances between themselves and the lead vehicle. Given that, for the drivers in the no anger-provocation groups, the distance between the lead vehicle and the driver’s vehicle is determined by approach speed and then remains relatively constant throughout the follow task, we did not examine the range variables for the drivers in these groups. Instead, we conducted a mixed ANOVA, again using a within subjects factor of phase (baseline/

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

1 It should be noted that data for drivers who crashed have been removed from the analysis and will be reflected in the altered degrees of freedom.

9781472414694_Dorn.indb 96

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:04 AM


maintenance) and a between group factor of PM errands (with/without) but this time on average and variation of speeds (km/h) during the un-provoking nonimpeding follow tasks. The ANOVA showed no main effect of phase on average speed and no interaction between PM errand and phase. All drivers maintained an average speed of approximately 40 km/h across both follow tasks, regardless of whether they were maintaining an errand intention or not. However, a main effect of phase approached significance on variation of speed (F(1,12) = 4.16, p = 0.06) with drivers having greater variation of speed during the PM maintenance phase (M = 12.35, SD = 2.59) than the baseline phase (M = 9.89, SD = 10.24). To explore this further, we conducted paired t-tests for each group comparing variation across baseline and the maintenance phases. The t-test approached significance (t(7) = −2.21, p = 0.06) for drivers with a PM intention to maintain, but not for the control group. The direction of the means suggest drivers with a PM errand intention had more speed variability during the maintenance phase (M = 12.85; SD = 2.5), than at baseline (M = 9.63, SD = 3.35). Drivers without PM errands had no difference between baseline (M = 10.24, SD = 3.50) and maintenance (M = 11.69, SD = 4.66). While these results are interesting from the perspective of identifying PM maintenance as a driving distractor, the ANOVAs failed to reach an acceptable significance level. Further, although we describe them as approaching significance it is unclear whether the p-value suggests something that is approaching or moving away from significance. Therefore, we must retain our original hypothesis that maintaining an intention to perform an errand was not sufficiently distracting to disrupt driver behaviour. Our finding that there was no difference in driving behaviours during the maintenance phase, regardless of PM task, supports the notion that maintenance of a delayed intention, as represented in this study, is resource neutral from a driving perspective. However, our findings clearly indicate the need for further studies to fully explore this issue. For example, although not statistically significant, the direction of means indicate that drivers with a PM task had to work harder to maintain consistent headway and speeds. Further data collection is currently underway on the study reported in this chapter. We posit that with a larger sample, the PM disruption may emerge and the relationship between anger and PM may also become clearer. Further research using more specific designs, such as increasing the number of errands or adding on-road hazards, may also uncover an effect of PM on driving performance. However, whether this putative effect would have functional significant is a different matter and also worthy of further research.

fC

op

y

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 H3: Drivers implementing a PM task will exhibit disrupted driving behaviour 39 compared to drivers without a PM task 40 A manipulation check on PM performance showed that drivers in the PM errand 41 groups correctly responded to an average of three out of the four target buildings. 42 Further, there was 100 per cent recall when tested about the tasks at the end of 43 44 the drive.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

97

9781472414694_Dorn.indb 97

5/30/2013 10:21:04 AM


98

oo

fC

op

y

In contrast to our expectations that maintaining an intention would not influence behaviour we expected that implementation would interfere with behaviour. More specifically that it would interrupt the ability to comply with the variable speed limits. We conducted two between group ANOVAs on average speed and percentage of time on correct speed across the implementation phase. The between subject variables were anger-provocation (impeded by lead driver/unimpeded by lead driver) and PM errands (with/without). We found a main effect of PM errand, with drivers required to search for specific buildings spending significantly less time at the correct speed limit (M = 12%, SD = 6%), when compared to drivers without the PM errands (M = 31%, SD = 15%; F(1,23) = 15.83, p < 0.001). Average speed also differed, with drivers implementing PM errands driving reliably slower (km/h) (M = 46.09, SD = 6.67) than drivers with no PM errands (M = 51.86, SD = 3.92; F(1,23) = 6.71, p < 0.05). Anger-provocation did not influence performance on variable speed limits and overall speed, neither were there interactions between PM and anger-provocation on these variables. Because self-reported anger subsided across the implementation drive, averaging performance across the entire 10 minute phase may wash out any initial effects of anger. To account for this we partitioned the drive into three segments of equal duration and re-ran the above analyses. We found no influence of angerprovocation on speed limit compliance or average speed in any of the segments (initial, middle, final). Arguably, by the time a PM errand can be implemented, the PM component is no different to a secondary visual search task. For example, having to scan the road-way for specific road-signage or monitor the driving display for changes in divided attention symbols. However, the disruption that emerged during the implementation phase strengthens our assumption that the intention to perform the PM errands was maintained during the maintenance phase. It is worth highlighting that drivers given PM errands seemed to prioritise these over speed limit compliance. The low percentage of time on correct speed limit suggests that drivers often ignored the variability of the speed limits. However, the placement of these signs was such that drivers would have to look past the speed limit signs to scan for target buildings. We have to stipulate, however, that whether these slower speeds would make drivers less safe is unclear.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Conclusions

We were unable to find a reliable interaction between self-reported anger and PM errands, and thus could not demonstrate that the PM task provides a distraction from anger or a reduction in anger-based behaviours. We did however find some tentative evidence to suggest that maintaining an intention to perform an errand can distract from the driving task. We conclude that these preliminary results highlight the need for further research into PM and driver behaviour.

9781472414694_Dorn.indb 98

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:04 AM


References

fC

op

y

PM provides a novel paradigm in which to examine driver distraction and inattention. The majority of the current driver inattention research has focused on identifying secondary events that distract the driver from the driving task. However, these all comprise physical activities that conflict with those required for safe driving (e.g., entering numbers on a mobile phone, or directing attention to a GPS display) or are related to shifts of attention caused by non-driving related stimuli (e.g., engaging in passenger conversation). Very little has been done to identify how internal self-generated cognitive distractions, such as maintaining an internal dialogue about an unrelated issue (e.g., remembering to stop at the bank on the way home), may interfere with the driving task. When this question has been examined, the findings suggest that certain cognitive tasks are less demanding than others. For example, solving trivia questions provides an adequate distraction from boredom states without impairing driving, whereas, working memory tasks do impair driving (Oron-Gilad et al., 2008). Savage, Potter and Tatler (2013) have also recently found that increased cognitive distraction (solving a puzzle) leads to slower and less accurate button-press responses to hazards in hazard perception videos. Thus, there is evidence to suggest PM errand maintenance may have a similar disruption as working memory tasks or puzzle solving. However, for now with our preliminary findings we conclude that more research should be undertaken to explore how simply maintaining an intention to do something at a later date may (or may not) impair driving performance. With a more detailed design and the inclusion of driving events that require a response from the driver, more obvious relationships between PM, anger and behaviour may become evident.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

99

Berdoulat, E., Vavassori, D., and Muñoz Sastre, M.T. (2013). Driving anger, emotional and instrumental aggressiveness, and impulsiveness in the prediction of aggressive and transgressive driving. Accident Analysis and Prevention, 50, 758–67. Brandamonte, M.A., Ferrante, D., Feresin, C., and Delbello, R. (2001). Dissociating prospective memory from vigilance processes. Psicologica, 22, 97–113. Bushman, B.J. (2002). Does venting anger feed or extinguish the flame? Catharsis, rumination, distraction, anger and aggressive responding. Personality and Social Psychology, 28, 724–31. Cheung, S.Y., and Lam, E.T.C. (2005). An innovative shorted bilingual version of the Profile of Mood States (POMS-SBV). School Psychology International, 26, 121–28. Einstein, G.O., McDaniel, M.A., Manzi, M., Cochran, B., and Baker, M. (2000). Prospective memory and aging: Forgetting intentions over short delays. Psychology and Aging, 15, 671–83.

9781472414694_Dorn.indb 99

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:04 AM


100

oo

fC

op

y

Einstein, G.O., McDaniel, M.A., Willford, C.L., Pagain, J.L., and Dismukes, R. (2003). Forgetting of intentions in demanding situations is rapid. Journal of Experimental Psychology: Applied, 9, 147–62. Einstein, G.O., and McDaniel, M.A. (2005). Prospective Memory Multiple Retrieval Processes. Current Directions in Psychological Science, 14, 286–90. Forgas, J.P. (1995). Mood and Judgment: The affect infusion model (AIM). Psychological Bulletin, 117(1), 39–66. Gerin, W., Davidson, K.W., Christenfeld, N.J.S., Goyl, T., and Swartz, J.E. (2006). The role of angry rumination and distraction in blood pressure recovery from emotional arousal. Psychosomatic Medicine, 68, 64–72. Lansdown, T.C., and Stephens, A.N. (2013). Couples, contentious conversations, mobile telephone use and driving. Accident Analysis and Prevention, 50, 416–22. Oron-Gilad, T., Ronen, A., and Shinar, D. (2008). Alertness maintaining tasks (AMTs) while driving. Accident Analysis and Prevention, 40, 851–60. Rendell, P.G., McDaniel, M.A., Forbes, R.D., and Einstein, G.O. (2007). Agerelated effects in prospective memory are modulated by ongoing task complexity and relation to target cue. Aging, Neuropsychology, and Cognition, 14, 236–56. Rusting, C.L., and Nolen-Hoeksema, S. (1998). Regulating responses to Anger: Effects of rumination and distraction on angry mood. Journal of Personality and Social Psychology, 74, 790–803. Savage, S.W., Potter, D.D., and Tatler, B.W. (2013). Does preoccupation impair hazard perception? A simultaneous EEG and eye tracking study. Transportation Research Part F, 17, 52–62. Stephens, A.N., and Groeger, J.A. (2011). Anger-congruent behaviour transfers across driving situations. Cognition and Emotion, 25, 1423–38. Stephens, A.N., Madigan, R., Trawley, S.L., and Groeger, J.A. (2013). Drivers display anger-congruent attention to potential traffic hazards. Applied Cognitive Psychology, DOI, 10.1002/acp.2894. Trawley, S.L., Rendell, P.G., Groeger, J.A., and Stephens, A.N. (under review). Prospective memory while driving: Comparison of time and event based intentions.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:04 AM

Chapter 8

Emotion Regulation of Car Drivers by the Physical and Psychological Parameters of Music Rainer Höger, Sabine Wollstädter, Sabine Eichhorst and Laura Becker

op

Introduction

y

Leuphana University of Lueneburg, Germany

fC

Driving takes place in a social environment where other road users have a strong influence on driving behaviour. One aspect of this social situation is interpersonal interaction which can lead to emotional responses in some cases. One of the most frequently emerging emotions in driving is that of anger (Nesbit et al., 2007) and the most common response to anger is aggressive behaviour. From the psychological point of view aggressive behaviour has a functional purpose. Aggressive behaviour is intended to restore the original unrestricted situation to allow free driving (Kaba et al., 1997, Shinar 1998, Neighbors et al., 2002). Aggressive driving, and in its extreme, road rage, are characterised by speeding, risky driving, and dangerous manoeuvres motivated by self-interest. Aggressive driving can be seen as a substantial contributor to traffic incidences. To understand the role of aggressive driving and its relationship with motor vehicle accidents, a survey was carried out by the American Automobile Association Foundation for Traffic Safety (2009). The results show that between 2003 and 2007 about 56 per cent of all motor vehicle accidents included at least one aggressive action of one of the participants. Thus emotions – especially anger – and their behavioural consequences play an important role in safe driving. If an emotional state, such as anger, contributes to accidents then the regulation of driver emotions can be a relevant accident prevention strategy. One possible way of influencing a driver’s emotional state consists of using the calming qualities of music in the vehicle. As shown by Wiesenthal et al. (2003) preferred music is able to calm down the driver, thereby reducing moderate levels of aggressive driving. Van der Zwaag et al. (2011) argue that music influences the driver’s mood, by lowering or raising mood levels which leads to different driving patterns. With reference to the dimensional approach of assessing emotions, two more general aspects are important for all emotions: valence and activation (Russell, 1980; Feldman, 1995).

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 101

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:04 AM


102

The fundamental idea for the present studies was to manipulate the impact of valence and activation of music in order to influence drivers’ emotional states. The first study used a quasi-experimental approach to investigate how valence could be associated with the process of anger regulation in certain driving situations, according to the emotional preference for different types of music. In the second study the musical tempo of preferred music was varied, in order to manipulate the music-induced activation. It was assumed that slow musical tempos would lead to a reduction in activation and therefore a decrease in anger experienced. Study 1 – Method

y

Participants

Material and procedure

op

A total of 43 participants took part in the study, as unpaid volunteers. The participants were aged between 18 and 37 years old and all participants held a current driving licence.

oo

fC

For the first step, participants were randomly assigned to one of two experimental conditions: (1) driving with preferred music, (2) driving with disliked music. Musical preferences for each participant were also collected. For the participants in the preferred music group the music to be played was selected based upon each individual’s preferences. For those in the disliked music group the music the individual most disliked was chosen to be played in the experiment. The musical styles chosen by the participants ranged from rock to classical music. The experimental setting consisted of a driving course in a driving simulator (STISIM). The driving simulator was equipped with three projectors which presented the traffic situation with a 180° field-of-view. To induce the emotional

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Figure 8.1

9781472414694_Dorn.indb 102

Anger inducing scenes A: traffic jam B: slow moving car C: tailgater

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:05 AM

Emotion Regulation by Music

y

Figure 8.2 Experimental order of events

fC

op

state of anger, traffic situations were created in which the driver was impeded during a driving session which had previously been relatively smooth and uninterrupted for a long period of time. Carefree driving was blocked or influenced by: (1) a slow moving car ahead, (2) a traffic jam, and (3) a vehicle tailgating the driver. The driving session started with a training phase followed by the three anger inducing situations (described above) which were presented in a random order for each of the participants. Between the anger inducing scenes there were periods of free unimpeded driving. Each anger inducing situation lasted for approximately four minutes. Within the first two minutes for each scene, the participant’s emotional state was recorded using a 7-point anger intensity scale. Then, depending on the experimental condition, preferred or disliked music was presented for two minutes at a comfortable sound level. Thus, within the anger inducing scenes two phases were presented: one without (t1, t3, t5) and one with music (t2, t4, t6). At the end of the musical episode the participants were required to judge their experienced anger intensity once more. The experiment employed a two factorial design with the factors type of music (preferred, disliked) and music presentation (with and without music).

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

103

Study 1: Results and Discussion

Firstly, participant’s judgements of the three anger inducing situations were aggregated so that mean values for anger intensity with music (t2, t4, t6) and without music (t1, t3, t5) were calculated. Figure 8.3 depicts the mean anger judgements for the experimental variables type of music and music presentation. The statistical analyses revealed a highly significant interaction between type of music and music presentation (F(1,41) = 3.92, p < 0.001, Eta2 = 0.46). Specifically, when preferred music was presented the level of anger reported was significantly lower than in the disliked music condition.

9781472414694_Dorn.indb 103

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:06 AM


op

fC

Figure 8.3 Mean rated intensity of anger, type of music and music presentation

oo

With reference to the dimensional approach to emotions, valence is concerned with internally represented stimuli and states (Russell, 1980). The quality of a certain piece of music and the quality of the emotional profile of anger have a common attribute of valence, in which both mental experiences can interact. Thus, the positive or negative valence of a piece of music interacts with the positive or negative valence of an emotional state. Whether an additive model is adequate for describing the processes has to be tested in further studies. In addition to the universal valence dimension, the activation dimension plays an important role in characterising the different states of emotions. Whether the activation level of music, as induced by the musical tempo, influences the intensity of anger experienced is the subject of Study 2. From the theoretical point of view a slow musical tempo should lead to a downsizing of the activation component of anger, whereas a high musical tempo is expected to increase the anger experienced.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

y

104

Study 2 – Method Participants A total of 57 participants took part in the study, as unpaid volunteers. The participants were aged between 18 and 40 years old and all participants held a current driving licence.

9781472414694_Dorn.indb 104

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:09 AM


Material and procedure

fC

op

y

As in Study 1, the second study also took place in a STISIM driving simulator. The same three events (slow moving car ahead, traffic jam, and vehicle tailgating the driver) were used to induce anger. Musical tempo was manipulated to investigate its effect on the level of activation. For each of the participants, their most preferred pieces of music were chosen to ensure that all musical pieces were comparable with respect to their individual valence. Then the participants were randomly assigned to one of three experimental conditions: (1) without any presented music (control), (2) with music of a low musical tempo, (3) with music of a high musical tempo. Furthermore, the preferred music of each participant was divided into those pieces with less than 90 beats per minute (low musical tempo) and those with more than 120 beats per minute (high musical tempo). Depending on the assigned condition, the participants either heard: no music, their preferred low tempo music or the preferred fast tempo music whilst driving. As in the first study, the driving session started with a training phase followed by the anger inducing situations, which were presented in a random order for each of the participants. Each of the single scenes lasted for about four minutes and for the first two minutes of each scene the emotional state of the individual was registered using a 7-point anger intensity scale. After these two minutes, one of three conditions was presented: no further music, slow paced music or fast paced music. At the end of the two minutes, the participants were asked to judge their experienced intensity of anger once again. This study used a two factor design with music tempo (no music, low, and high) and a repeated measure factor music presentation (with, without).

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

105

Study 2: Results and Discussion

The reported anger intensity measures were aggregated over the three anger inducing situations. Following this, anger intensity measures at the second measurement point (music presentation, excluding the control group) were subtracted from the measures of the first measurement time period (no music presentation) in order to measure the amount of anger regulation for two music groups (low tempo, high tempo) and the control group. The following graph shows the mean values for the created variable anger regulation for the control group and the two groups in the music presentation conditions (Figure 8.4 below). As shown in the graph the mean value of anger regulation in the control group was near zero, as expected. However, anger regulation was highest for the group with the low tempo music, while for the high tempo group anger regulation was between the two other conditions. An ANOVA was conducted on the data, which revealed a significant main effect for the three different experimental conditions (F(2,54) = 4.21, p < 0.05, Eta2 = 0.14). Post hoc analyses, using the Scheffé procedure, were carried out in order to test for differences

9781472414694_Dorn.indb 105

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:09 AM


op

fC

Figure 8.4 Amount of anger reduction in relation to the experimental conditions

oo

between the three groups. This showed that the only significant difference was between the control group (no music) and the group where the preferred music was presented with a low tempo ( p < 0.05). The group with the fast tempo music was not significantly different to the control group or to the slow music group. The results of the second study again showed the influence of music on the regulation of anger experienced in a driving situation. This time the latent dimension, activation, was manipulated using the tempo of the music. However, although as expected the music with a low tempo reduced the effect of anger, the high tempo music was not significantly different from the condition where no music was presented. Furthermore, the effect of the fast tempo music was also not significantly different from that of slow music. Thus, it is not clear whether the high tempo music used here was not strong enough to elevate the emotional activation or whether the effect of fast music was compensated by the positive valence of the music. The latter means that activation and valence can interact to result in a particular manifestation of an emotion.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

y

106

General Discussion Inducting emotions within an artificial experimental settings is not easy (Roidl et al., 2012), but the findings from the present research suggest that it is possible to do so using a driving simulator. Occasionally participants were even observed pressing the horn in the anger inducing traffic scenes.

9781472414694_Dorn.indb 106

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:10 AM


fC

op

y

The results from both studies suggest that valence and activation using music are able to influence the intensity of anger. From a theoretical point of view it is argued that experienced music and experienced anger are manifestations of the dimensions of valence and activation which interact and lead to a modification of the emotional state. However, the manner in which these two variables interact will require further research. Another explanation for the findings might be that the music causes a shift of attention away from the emotional state. Evidence to support this explanation comes from Wollstädter et al. (2012) who conducted an online study in which drivers were asked about how they regulated their emotions during driving. This research found that music distracts from an event which might otherwise cause a negative emotion. Thus, the higher the valence and activation components of a particular piece of music, the more attention will be drawn away by the music. From an applied perspective an important question is, whether or not music can be used to increase safe driving behaviour? However, the two performed simulator studies on this topic reported inconsistent findings. As shown by other research groups the impact of music on driving behaviour depends on the respective framework conditions. Van der Zwaag et al. (2012) reported that under driving situations with low demand, driving speed was affected by music the individual liked. Participants lowered their speed and drove more carefully, compared to the situation with disliked music. On the other hand, Pêcher et al. (2009) found a deterioration of driving performance when happy music was presented. Therefore, a series of additional variables may need to be taken into account, including the level of demand imposed by the driving task, the emotions involved and the manifestations of the latent dimensions of valence and activation of the music.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

107

Acknowledgements

This research was supported by a grant (F.A.-Nr. 2006.63) from the ‘Arbeitsgemeinschaft Innovative Projekte’ (AGIP) of the Ministry of Science and Culture, Lower Saxony, Germany. References

American Automobile Association Foundation for Traffic Safety (2009). Aggressive driving – research update. Washington, DC: AAA foundation for Traffic Safety. Feldman, L.A. (1995). Variations in the circumplex structure of mood. Personality and Social Psychology Bulletin, 21, 806–17. Kaba, A., Ebenbichler, R., and Brandstätter, C. (1997). Aggression in road traffic (Research paper). [Aggression im Straßenverkehr (Projektbericht)]. Wien, Austria: Institut für Verkehrspsychologie.

9781472414694_Dorn.indb 107

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:10 AM


108

oo

fC

op

y

Neighbors, C., Victor, N.A., and Knee, C.R. (2002). A motivational model of driving anger and aggression. Personality and Social Psychology Bulletin, 28, 324–35. Nesbit, S.M., Conger, J.C., and Conger, A.J. (2007). A quantitative review of the relationship between anger and aggressive driving. Aggression and Violent Behavior, 12, 156–76. Pêcher, C., Lemercier, C., and Cellier, J.-M. (2009). Emotions drive attention: Effects on driver’s behaviour. Safety Science, 47, 1254–59. Roidl, E., Frehse, B., and Höger, R. (2013). Emotional states of drivers and the impact on driving behaviour. Accident Analysis and Prevention (submitted). Russell, J.A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39, 1161–78. Shinar, D. (1998). Aggressive driving: An observational study of driver, vehicle, and situational variables. Accident Analysis and Prevention, 36, 429–37. van der Zwaag, M.D., Fairclough, S., Spiridon, E., and Westerink. J.H.D.M. (2011). The impact of music on affect during anger inducing drives. In S. D’Mello, A. Graesser, B. Schuller, and J.C. Martin (eds), Affective computing and intelligent interaction (pp. 407–16). Lecture notes in Computer Science. Berlin: Springer. van der Zwaag, M.D., Dijksterhus, C., de Ward, D., Mulder B.L.J.M., Westerink, J.H.D.M., and Brookhuis, K.A. (2012). The influence of music on mood and performance while driving. Ergonomics, 55, 12–22. Wiesenthal, D.L., Hennessy, D.A., and Totten, B. (2003). The influence of music on mild driver aggression. Transportation Research Part F: Traffic Psychology and Behaviour, 6, 125–34. Wollstädter, S., Pfister, R., Vollrath, M., and Höger, R. (2012). Don’t be upset! Can cars regulate anger by communication? In D. de Waard, N. Merat, H. Jamson, Y. Banard, and O. Carsten (eds), Human factors of systems and technology (pp. 223–34). Maastricht, Netherlands: Shaker Publishing.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 108

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:10 AM

Chapter 9

The Relationship between Seat Belt Use and Distracted Driving M. Eugènia Gras*, Francesc Prat*, Montserrat Planes*, Sílvia Font-Mayolas* and Mark J.M. Sullman**

Introduction

y

University of Girona, Spain; ** Cranfield University, UK

op

*

fC

Driving is a complex activity which requires the driver to be in perfect psychophysical condition in order for them to be able to cope with the ever changing traffic environment (Gras, Planes and Font-Mayolas, 2008). However, despite the complexity of the traffic environment, many drivers feel capable of carrying out secondary activities at the same time as driving. In Spain, research has shown that the most common of these secondary activities are: talking to a passenger, smoking, using a mobile phone and adjusting the stereo (Gras, Planes, Font-Mayolas, Sullman, Jiménez and Prat, 2012; RACC, 2006). Not surprisingly, there is extensive evidence that engaging in a secondary activity while driving increases the likelihood of having a road traffic accident (Klauer, Dingus, Neale, Sudweeks and Ramsey, 2006; Stutts, Reinfurt, Staplin and Rodgman, 2001). Specifically, in Spain it has been estimated that 37 per cent of motor vehicle crashes are caused by driver distraction (WHO, 2011). If a distracted driver does crash it is important that they are wearing a seat belt, as the seriousness of injuries increases greatly if the occupants are not wearing a seat belt. Although the vast majority of drivers in Spain use a seat belt, a small but significant proportion of drivers do not use it consistently (IRTAD, 2011). Moreover, the rate of seat belt use in Spain varies according to the type of road driven on, in other words whether they were travelling on a motorway (roads between cities) or an urban road (roads within cities or towns) (Gras, Cunill, Sullman, Planes and Font-Mayolas, 2007). Gras et al. (2007) found that one of the best predictors of observed seat belt use was the type of road travelled: the frequency of use was lower amongst drivers who reported having travelled solely on urban roads, compared with those who had also travelled on motorways. There are a number of explanations for the differences by road type. Firstly, seat belt use on the motorway was compulsory in Spain 20 years before it was compulsory on urban roads. This meant that for a long time drivers were not used to wearing a seat belt within cities or towns. It is also well known that the

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 109

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:10 AM


110

oo

fC

op

y

older the compulsory law for using a seat belt, the higher the usage rate (IRTAD, 2011). In addition, as the speed limit in urban roads is considerably lower (between 30 and 50 km/h) than on the motorways (between 80 and 120 km/h), drivers tend to perceive it to be more risky to not use a seat belt on motorways than while driving on urban roads (Gras et al., 2007). Furthermore, the level of enforcement for not using a seat belt on Spanish urban roads has traditionally been lower than on the motorways. However, although the implementation of the penalty point law in 2006 increased seat belt usage on both types of roads, the differences between them still remains (Gras, Font-Mayolas, Salamó, Cunill, and Planes, 2010; IRTAD, 2011). For example, in 2010 Spanish drivers’ seat belt wearing rates were found to be 95 per cent outside urban areas and 83 per cent inside urban areas (IRTAD, 2011). The results of recent research have found that drivers who reported more frequently using a mobile phone while driving also drove more riskily (e.g., drove faster and changed lanes more frequently) than those who reported that they never or rarely use a mobile phone while driving (Zhao, Reimer, Mehler, Ambrosio and Coughlin, 2012). In addition, these authors found that the attitude towards risky driving behaviour was more permissive amongst the more frequent mobile phone users and these drivers also had higher scores on the violations subscale of the Driving Behaviour Questionnaire. Similar results were also found by Hamilton, Arnold and Tefft (2013) with regards to speed limits: drivers who reported a greater frequency of mobile phone use while driving were also more likely to report speeding. Moreover, previous observational research has found a relationship between using a mobile phone while driving and seat belt use. Specifically, drivers who were observed using a handheld mobile phone while driving were also much more likely to be observed not using a seat belt, than drivers who were not using a mobile phone (Eby and Vivoda, 2003; Narine, Walter and Charman, 2009; Wilson, Fang, Wiggins and Cooper, 2003). Similar results have also been found using self-report questionnaires (Hamilton et al., 2013; Seo and Torabi, 2004). These results suggest that, overall, some drivers are more likely to engage in a range of different types of risky behaviour while driving or are more risk prone than others (Hamilton et al., 2013). The research regarding the relationship between seat belt use and distracted driving has thus far solely focused on mobile phone use. However, currently no research has investigated the relationship between seat belt use and other types of secondary activities that might distract drivers. Therefore, the present study investigated the prevalence of seat belt use while driving on urban roads, by gender, age and whether they were also observed engaged in a secondary activity while driving.

Pr

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

9781472414694_Dorn.indb 110

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:10 AM

The Relationship between Seat Belt Use and Distracted Driving

Method Material and procedure

fC

op

y

A cross-sectional observational study was carried out in the city of Girona (Spain) in the spring of 2011. Observations were made over 63 daylight hours and included 6,578 drivers. The roads where observations would take place were randomly selected from an exhaustive list of passable roads obtained from the Girona City Council. Two independent observers recorded: gender, estimated age range (under 30, between 30 and 50, over 50), whether the drivers were wearing a seat belt and also whether they were driving only or were also engaged in some type of visible secondary activity while driving. The definitions of the secondary activities were developed prior to conducting the observations and included: using a mobile phone (they were clearly holding a mobile phone in their hand or to one ear), other technological device (they were manipulating any other device such as the heater, the GPS, the radio or the CD player), drinking or eating (they were holding or eating/drinking something in a clear and visible manner), smoking (they were holding a cigarette in their mouth or in one hand), talking to a passenger (the driver’s lips were moving and/or gesticulating – only if there were one or more passengers in the car) or other activity (any other visible activity which did not fit into one of the previous categories). Drivers were classified into two groups: those who were only driving and those who were engaged in at least one visible secondary activity. Because of the high density of traffic on the roads not all the drivers could be observed. Therefore, the selection of the vehicles to be observed was made using Gras et al.’s (2012) procedure. Specifically, prior to beginning each session observers activated a timing device. The timer rang ten seconds after activation and the observers recorded the data from the first car to arrive. When two cars arrived at the same time, the observers selected the one in the closest lane. Emergency vehicles such as ambulances, driving school cars and marked police cars as well as heavy and light trucks, motorcycles and buses were excluded. Cohen’s Kappa Coefficients were computed to assess the level of agreement between observers. Cohen’s Kappa for seat belt use was 0.94, while for secondary activities these coefficients ranged from 0.70 (using other technological device) to 0.96 (using a mobile phone). For gender and estimated age the Kappa coefficients were 0.97 and 0.67, respectively. Coefficients over 0.60 are generally regarded as being indicative of a good agreement, while above 0.81 reflects very good agreement (Altman, 1991). Data from one observer was then randomly selected and analysed using SPSS version 19. Chi-square tests were performed to analyse the relationships between the variables.

Pr oo

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

111

9781472414694_Dorn.indb 111

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

5/30/2013 10:21:10 AM


112

Results Almost seven out of ten drivers observed were males (68.7 per cent). The driver’s distribution by estimated age was: 15.8 per cent under 30 years old, 50.5 per cent between 30 and 49 years old, and 33.7 per cent over 50 years old. Overall, 7 per cent of the observed drivers were not wearing a seat belt. Table 9.1 presents the percentage of drivers observed wearing a seat belt by estimated age and gender. Significantly more men than women were not wearing a seat belt (χ 2(1) = 39.6; p < 0.0005) and significantly more older drivers (estimated age over 50) were observed not wearing a seat belt (χ2(2) = 8.3; p = 0.016), compared with younger drivers.

Male Female

Wearing a seat belt

Not wearing a seat belt

91.6

8.4

95.9

4.1

94.0

6.0

30–50 years

93.5

6.5

>50 years

91.7

8.3

fC